ZhangYunchenY
commited on
Commit
·
fff6bd9
1
Parent(s):
a51520a
[Model] bart-base-qqp
Browse files- config.json +74 -0
- merges.txt +0 -0
- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +1 -0
- tokenizer.json +0 -0
- tokenizer_config.json +1 -0
- trainer_state.json +3316 -0
- training_args.bin +3 -0
- vocab.json +0 -0
config.json
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/mnt/lustre/zhangyunchen/transformers/bart-base",
|
| 3 |
+
"activation_dropout": 0.1,
|
| 4 |
+
"activation_function": "gelu",
|
| 5 |
+
"add_bias_logits": false,
|
| 6 |
+
"add_final_layer_norm": false,
|
| 7 |
+
"architectures": [
|
| 8 |
+
"BartForSequenceClassification"
|
| 9 |
+
],
|
| 10 |
+
"attention_dropout": 0.1,
|
| 11 |
+
"bos_token_id": 0,
|
| 12 |
+
"classif_dropout": 0.1,
|
| 13 |
+
"classifier_dropout": 0.0,
|
| 14 |
+
"d_model": 768,
|
| 15 |
+
"decoder_attention_heads": 12,
|
| 16 |
+
"decoder_ffn_dim": 3072,
|
| 17 |
+
"decoder_layerdrop": 0.0,
|
| 18 |
+
"decoder_layers": 6,
|
| 19 |
+
"decoder_start_token_id": 2,
|
| 20 |
+
"dropout": 0.1,
|
| 21 |
+
"early_stopping": true,
|
| 22 |
+
"encoder_attention_heads": 12,
|
| 23 |
+
"encoder_ffn_dim": 3072,
|
| 24 |
+
"encoder_layerdrop": 0.0,
|
| 25 |
+
"encoder_layers": 6,
|
| 26 |
+
"eos_token_id": 2,
|
| 27 |
+
"finetuning_task": "qqp",
|
| 28 |
+
"forced_eos_token_id": 2,
|
| 29 |
+
"gradient_checkpointing": false,
|
| 30 |
+
"id2label": {
|
| 31 |
+
"0": "not_duplicate",
|
| 32 |
+
"1": "duplicate"
|
| 33 |
+
},
|
| 34 |
+
"init_std": 0.02,
|
| 35 |
+
"is_encoder_decoder": true,
|
| 36 |
+
"label2id": {
|
| 37 |
+
"duplicate": 1,
|
| 38 |
+
"not_duplicate": 0
|
| 39 |
+
},
|
| 40 |
+
"max_position_embeddings": 1024,
|
| 41 |
+
"model_type": "bart",
|
| 42 |
+
"no_repeat_ngram_size": 3,
|
| 43 |
+
"normalize_before": false,
|
| 44 |
+
"normalize_embedding": true,
|
| 45 |
+
"num_beams": 4,
|
| 46 |
+
"num_hidden_layers": 6,
|
| 47 |
+
"pad_token_id": 1,
|
| 48 |
+
"problem_type": "single_label_classification",
|
| 49 |
+
"scale_embedding": false,
|
| 50 |
+
"task_specific_params": {
|
| 51 |
+
"summarization": {
|
| 52 |
+
"length_penalty": 1.0,
|
| 53 |
+
"max_length": 128,
|
| 54 |
+
"min_length": 12,
|
| 55 |
+
"num_beams": 4
|
| 56 |
+
},
|
| 57 |
+
"summarization_cnn": {
|
| 58 |
+
"length_penalty": 2.0,
|
| 59 |
+
"max_length": 142,
|
| 60 |
+
"min_length": 56,
|
| 61 |
+
"num_beams": 4
|
| 62 |
+
},
|
| 63 |
+
"summarization_xsum": {
|
| 64 |
+
"length_penalty": 1.0,
|
| 65 |
+
"max_length": 62,
|
| 66 |
+
"min_length": 11,
|
| 67 |
+
"num_beams": 6
|
| 68 |
+
}
|
| 69 |
+
},
|
| 70 |
+
"torch_dtype": "float32",
|
| 71 |
+
"transformers_version": "4.16.2",
|
| 72 |
+
"use_cache": true,
|
| 73 |
+
"vocab_size": 50265
|
| 74 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2707878876f58372c902d3bb32adefb48b0e8f46c9565ffe67edcb36c343fd89
|
| 3 |
+
size 1120267003
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58c3fdbafdc9e00d90d9037f23346b6edcc5ac6bb56addbe9649a3e43adef004
|
| 3 |
+
size 560154805
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a36bf66c9d9314a6035a9c8f3a9c6d17f6af6077e09f7fc342e8c5ddcba4126b
|
| 3 |
+
size 14659
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ea41e47aed62dd32acf53b1b88cb8383a89c1377c751e727dca4422d294a8b0d
|
| 3 |
+
size 623
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "add_prefix_space": false, "errors": "replace", "sep_token": "</s>", "cls_token": "<s>", "pad_token": "<pad>", "mask_token": "<mask>", "trim_offsets": true, "special_tokens_map_file": null, "name_or_path": "/mnt/lustre/zhangyunchen/transformers/bart-base", "tokenizer_class": "BartTokenizer"}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3316 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 9.67373142203852,
|
| 5 |
+
"global_step": 110000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"learning_rate": 5.885815185403178e-07,
|
| 13 |
+
"loss": 0.6708,
|
| 14 |
+
"step": 200
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.04,
|
| 18 |
+
"learning_rate": 1.1771630370806356e-06,
|
| 19 |
+
"loss": 0.6457,
|
| 20 |
+
"step": 400
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.05,
|
| 24 |
+
"learning_rate": 1.7657445556209538e-06,
|
| 25 |
+
"loss": 0.6149,
|
| 26 |
+
"step": 600
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.07,
|
| 30 |
+
"learning_rate": 2.3543260741612712e-06,
|
| 31 |
+
"loss": 0.5516,
|
| 32 |
+
"step": 800
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.09,
|
| 36 |
+
"learning_rate": 2.942907592701589e-06,
|
| 37 |
+
"loss": 0.5014,
|
| 38 |
+
"step": 1000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.11,
|
| 42 |
+
"learning_rate": 3.5314891112419075e-06,
|
| 43 |
+
"loss": 0.484,
|
| 44 |
+
"step": 1200
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12,
|
| 48 |
+
"learning_rate": 4.120070629782225e-06,
|
| 49 |
+
"loss": 0.4428,
|
| 50 |
+
"step": 1400
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.14,
|
| 54 |
+
"learning_rate": 4.7086521483225425e-06,
|
| 55 |
+
"loss": 0.4323,
|
| 56 |
+
"step": 1600
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.16,
|
| 60 |
+
"learning_rate": 5.297233666862861e-06,
|
| 61 |
+
"loss": 0.4234,
|
| 62 |
+
"step": 1800
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.18,
|
| 66 |
+
"learning_rate": 5.885815185403178e-06,
|
| 67 |
+
"loss": 0.402,
|
| 68 |
+
"step": 2000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.19,
|
| 72 |
+
"learning_rate": 6.474396703943496e-06,
|
| 73 |
+
"loss": 0.4142,
|
| 74 |
+
"step": 2200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.21,
|
| 78 |
+
"learning_rate": 7.062978222483815e-06,
|
| 79 |
+
"loss": 0.39,
|
| 80 |
+
"step": 2400
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.23,
|
| 84 |
+
"learning_rate": 7.651559741024132e-06,
|
| 85 |
+
"loss": 0.3842,
|
| 86 |
+
"step": 2600
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.25,
|
| 90 |
+
"learning_rate": 8.24014125956445e-06,
|
| 91 |
+
"loss": 0.3747,
|
| 92 |
+
"step": 2800
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.26,
|
| 96 |
+
"learning_rate": 8.828722778104768e-06,
|
| 97 |
+
"loss": 0.37,
|
| 98 |
+
"step": 3000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.28,
|
| 102 |
+
"learning_rate": 9.417304296645085e-06,
|
| 103 |
+
"loss": 0.3688,
|
| 104 |
+
"step": 3200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.3,
|
| 108 |
+
"learning_rate": 1.0005885815185404e-05,
|
| 109 |
+
"loss": 0.3656,
|
| 110 |
+
"step": 3400
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.32,
|
| 114 |
+
"learning_rate": 1.0594467333725723e-05,
|
| 115 |
+
"loss": 0.3488,
|
| 116 |
+
"step": 3600
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.33,
|
| 120 |
+
"learning_rate": 1.1183048852266041e-05,
|
| 121 |
+
"loss": 0.3532,
|
| 122 |
+
"step": 3800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.35,
|
| 126 |
+
"learning_rate": 1.1771630370806357e-05,
|
| 127 |
+
"loss": 0.3553,
|
| 128 |
+
"step": 4000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.37,
|
| 132 |
+
"learning_rate": 1.2360211889346675e-05,
|
| 133 |
+
"loss": 0.3597,
|
| 134 |
+
"step": 4200
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.39,
|
| 138 |
+
"learning_rate": 1.2948793407886992e-05,
|
| 139 |
+
"loss": 0.3427,
|
| 140 |
+
"step": 4400
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.4,
|
| 144 |
+
"learning_rate": 1.3537374926427311e-05,
|
| 145 |
+
"loss": 0.35,
|
| 146 |
+
"step": 4600
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.42,
|
| 150 |
+
"learning_rate": 1.412595644496763e-05,
|
| 151 |
+
"loss": 0.342,
|
| 152 |
+
"step": 4800
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.44,
|
| 156 |
+
"learning_rate": 1.4714537963507947e-05,
|
| 157 |
+
"loss": 0.3304,
|
| 158 |
+
"step": 5000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.46,
|
| 162 |
+
"learning_rate": 1.5303119482048264e-05,
|
| 163 |
+
"loss": 0.3326,
|
| 164 |
+
"step": 5200
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.47,
|
| 168 |
+
"learning_rate": 1.589170100058858e-05,
|
| 169 |
+
"loss": 0.3273,
|
| 170 |
+
"step": 5400
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.49,
|
| 174 |
+
"learning_rate": 1.64802825191289e-05,
|
| 175 |
+
"loss": 0.3226,
|
| 176 |
+
"step": 5600
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.51,
|
| 180 |
+
"learning_rate": 1.706886403766922e-05,
|
| 181 |
+
"loss": 0.329,
|
| 182 |
+
"step": 5800
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.53,
|
| 186 |
+
"learning_rate": 1.7657445556209536e-05,
|
| 187 |
+
"loss": 0.3096,
|
| 188 |
+
"step": 6000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.55,
|
| 192 |
+
"learning_rate": 1.8246027074749856e-05,
|
| 193 |
+
"loss": 0.3196,
|
| 194 |
+
"step": 6200
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.56,
|
| 198 |
+
"learning_rate": 1.883460859329017e-05,
|
| 199 |
+
"loss": 0.3228,
|
| 200 |
+
"step": 6400
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.58,
|
| 204 |
+
"learning_rate": 1.942319011183049e-05,
|
| 205 |
+
"loss": 0.3164,
|
| 206 |
+
"step": 6600
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.6,
|
| 210 |
+
"learning_rate": 1.999925173503938e-05,
|
| 211 |
+
"loss": 0.2999,
|
| 212 |
+
"step": 6800
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.62,
|
| 216 |
+
"learning_rate": 1.996183848700825e-05,
|
| 217 |
+
"loss": 0.3205,
|
| 218 |
+
"step": 7000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.63,
|
| 222 |
+
"learning_rate": 1.9924425238977122e-05,
|
| 223 |
+
"loss": 0.3053,
|
| 224 |
+
"step": 7200
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.65,
|
| 228 |
+
"learning_rate": 1.9887011990945994e-05,
|
| 229 |
+
"loss": 0.2995,
|
| 230 |
+
"step": 7400
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.67,
|
| 234 |
+
"learning_rate": 1.984959874291487e-05,
|
| 235 |
+
"loss": 0.313,
|
| 236 |
+
"step": 7600
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.69,
|
| 240 |
+
"learning_rate": 1.981218549488374e-05,
|
| 241 |
+
"loss": 0.2996,
|
| 242 |
+
"step": 7800
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.7,
|
| 246 |
+
"learning_rate": 1.977477224685261e-05,
|
| 247 |
+
"loss": 0.3289,
|
| 248 |
+
"step": 8000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.72,
|
| 252 |
+
"learning_rate": 1.9737358998821485e-05,
|
| 253 |
+
"loss": 0.3015,
|
| 254 |
+
"step": 8200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.74,
|
| 258 |
+
"learning_rate": 1.9699945750790357e-05,
|
| 259 |
+
"loss": 0.3029,
|
| 260 |
+
"step": 8400
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.76,
|
| 264 |
+
"learning_rate": 1.966253250275923e-05,
|
| 265 |
+
"loss": 0.3075,
|
| 266 |
+
"step": 8600
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.77,
|
| 270 |
+
"learning_rate": 1.96251192547281e-05,
|
| 271 |
+
"loss": 0.2972,
|
| 272 |
+
"step": 8800
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.79,
|
| 276 |
+
"learning_rate": 1.9587706006696973e-05,
|
| 277 |
+
"loss": 0.2984,
|
| 278 |
+
"step": 9000
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.81,
|
| 282 |
+
"learning_rate": 1.9550292758665845e-05,
|
| 283 |
+
"loss": 0.2894,
|
| 284 |
+
"step": 9200
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.83,
|
| 288 |
+
"learning_rate": 1.9512879510634717e-05,
|
| 289 |
+
"loss": 0.2857,
|
| 290 |
+
"step": 9400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.84,
|
| 294 |
+
"learning_rate": 1.9475466262603592e-05,
|
| 295 |
+
"loss": 0.2812,
|
| 296 |
+
"step": 9600
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.86,
|
| 300 |
+
"learning_rate": 1.943805301457246e-05,
|
| 301 |
+
"loss": 0.2872,
|
| 302 |
+
"step": 9800
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.88,
|
| 306 |
+
"learning_rate": 1.9400639766541333e-05,
|
| 307 |
+
"loss": 0.289,
|
| 308 |
+
"step": 10000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.9,
|
| 312 |
+
"learning_rate": 1.9363226518510205e-05,
|
| 313 |
+
"loss": 0.2791,
|
| 314 |
+
"step": 10200
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.91,
|
| 318 |
+
"learning_rate": 1.932581327047908e-05,
|
| 319 |
+
"loss": 0.2912,
|
| 320 |
+
"step": 10400
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.93,
|
| 324 |
+
"learning_rate": 1.9288400022447952e-05,
|
| 325 |
+
"loss": 0.2872,
|
| 326 |
+
"step": 10600
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.95,
|
| 330 |
+
"learning_rate": 1.9250986774416824e-05,
|
| 331 |
+
"loss": 0.2766,
|
| 332 |
+
"step": 10800
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.97,
|
| 336 |
+
"learning_rate": 1.9213573526385692e-05,
|
| 337 |
+
"loss": 0.2819,
|
| 338 |
+
"step": 11000
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.98,
|
| 342 |
+
"learning_rate": 1.9176160278354568e-05,
|
| 343 |
+
"loss": 0.2795,
|
| 344 |
+
"step": 11200
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.0,
|
| 348 |
+
"learning_rate": 1.913874703032344e-05,
|
| 349 |
+
"loss": 0.2677,
|
| 350 |
+
"step": 11400
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.02,
|
| 354 |
+
"learning_rate": 1.910133378229231e-05,
|
| 355 |
+
"loss": 0.2488,
|
| 356 |
+
"step": 11600
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 1.04,
|
| 360 |
+
"learning_rate": 1.9063920534261183e-05,
|
| 361 |
+
"loss": 0.256,
|
| 362 |
+
"step": 11800
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 1.06,
|
| 366 |
+
"learning_rate": 1.9026507286230055e-05,
|
| 367 |
+
"loss": 0.2489,
|
| 368 |
+
"step": 12000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 1.07,
|
| 372 |
+
"learning_rate": 1.8989094038198927e-05,
|
| 373 |
+
"loss": 0.2525,
|
| 374 |
+
"step": 12200
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.09,
|
| 378 |
+
"learning_rate": 1.89516807901678e-05,
|
| 379 |
+
"loss": 0.2483,
|
| 380 |
+
"step": 12400
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 1.11,
|
| 384 |
+
"learning_rate": 1.8914267542136674e-05,
|
| 385 |
+
"loss": 0.2528,
|
| 386 |
+
"step": 12600
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.13,
|
| 390 |
+
"learning_rate": 1.8876854294105543e-05,
|
| 391 |
+
"loss": 0.2481,
|
| 392 |
+
"step": 12800
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 1.14,
|
| 396 |
+
"learning_rate": 1.8839441046074415e-05,
|
| 397 |
+
"loss": 0.2517,
|
| 398 |
+
"step": 13000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 1.16,
|
| 402 |
+
"learning_rate": 1.880202779804329e-05,
|
| 403 |
+
"loss": 0.2514,
|
| 404 |
+
"step": 13200
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.18,
|
| 408 |
+
"learning_rate": 1.8764614550012162e-05,
|
| 409 |
+
"loss": 0.2464,
|
| 410 |
+
"step": 13400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.2,
|
| 414 |
+
"learning_rate": 1.8727201301981034e-05,
|
| 415 |
+
"loss": 0.2586,
|
| 416 |
+
"step": 13600
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.21,
|
| 420 |
+
"learning_rate": 1.8689788053949906e-05,
|
| 421 |
+
"loss": 0.2507,
|
| 422 |
+
"step": 13800
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.23,
|
| 426 |
+
"learning_rate": 1.8652374805918778e-05,
|
| 427 |
+
"loss": 0.2609,
|
| 428 |
+
"step": 14000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 1.25,
|
| 432 |
+
"learning_rate": 1.861496155788765e-05,
|
| 433 |
+
"loss": 0.2368,
|
| 434 |
+
"step": 14200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.27,
|
| 438 |
+
"learning_rate": 1.857754830985652e-05,
|
| 439 |
+
"loss": 0.2473,
|
| 440 |
+
"step": 14400
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.28,
|
| 444 |
+
"learning_rate": 1.8540135061825394e-05,
|
| 445 |
+
"loss": 0.2379,
|
| 446 |
+
"step": 14600
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.3,
|
| 450 |
+
"learning_rate": 1.8502721813794265e-05,
|
| 451 |
+
"loss": 0.2431,
|
| 452 |
+
"step": 14800
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.32,
|
| 456 |
+
"learning_rate": 1.8465308565763137e-05,
|
| 457 |
+
"loss": 0.2521,
|
| 458 |
+
"step": 15000
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.34,
|
| 462 |
+
"learning_rate": 1.842789531773201e-05,
|
| 463 |
+
"loss": 0.2473,
|
| 464 |
+
"step": 15200
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.35,
|
| 468 |
+
"learning_rate": 1.8390482069700885e-05,
|
| 469 |
+
"loss": 0.2496,
|
| 470 |
+
"step": 15400
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.37,
|
| 474 |
+
"learning_rate": 1.8353068821669757e-05,
|
| 475 |
+
"loss": 0.2529,
|
| 476 |
+
"step": 15600
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.39,
|
| 480 |
+
"learning_rate": 1.8315655573638625e-05,
|
| 481 |
+
"loss": 0.2484,
|
| 482 |
+
"step": 15800
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.41,
|
| 486 |
+
"learning_rate": 1.8278242325607497e-05,
|
| 487 |
+
"loss": 0.2393,
|
| 488 |
+
"step": 16000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 1.42,
|
| 492 |
+
"learning_rate": 1.8240829077576372e-05,
|
| 493 |
+
"loss": 0.2394,
|
| 494 |
+
"step": 16200
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.44,
|
| 498 |
+
"learning_rate": 1.8203415829545244e-05,
|
| 499 |
+
"loss": 0.2265,
|
| 500 |
+
"step": 16400
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.46,
|
| 504 |
+
"learning_rate": 1.8166002581514116e-05,
|
| 505 |
+
"loss": 0.2435,
|
| 506 |
+
"step": 16600
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.48,
|
| 510 |
+
"learning_rate": 1.8128589333482988e-05,
|
| 511 |
+
"loss": 0.2513,
|
| 512 |
+
"step": 16800
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 1.5,
|
| 516 |
+
"learning_rate": 1.809117608545186e-05,
|
| 517 |
+
"loss": 0.2478,
|
| 518 |
+
"step": 17000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 1.51,
|
| 522 |
+
"learning_rate": 1.8053762837420732e-05,
|
| 523 |
+
"loss": 0.2601,
|
| 524 |
+
"step": 17200
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.53,
|
| 528 |
+
"learning_rate": 1.8016349589389604e-05,
|
| 529 |
+
"loss": 0.2371,
|
| 530 |
+
"step": 17400
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 1.55,
|
| 534 |
+
"learning_rate": 1.7978936341358476e-05,
|
| 535 |
+
"loss": 0.2504,
|
| 536 |
+
"step": 17600
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.57,
|
| 540 |
+
"learning_rate": 1.7941523093327348e-05,
|
| 541 |
+
"loss": 0.2414,
|
| 542 |
+
"step": 17800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.58,
|
| 546 |
+
"learning_rate": 1.790410984529622e-05,
|
| 547 |
+
"loss": 0.2296,
|
| 548 |
+
"step": 18000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 1.6,
|
| 552 |
+
"learning_rate": 1.7866696597265095e-05,
|
| 553 |
+
"loss": 0.2413,
|
| 554 |
+
"step": 18200
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 1.62,
|
| 558 |
+
"learning_rate": 1.7829283349233967e-05,
|
| 559 |
+
"loss": 0.2447,
|
| 560 |
+
"step": 18400
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.64,
|
| 564 |
+
"learning_rate": 1.779187010120284e-05,
|
| 565 |
+
"loss": 0.2392,
|
| 566 |
+
"step": 18600
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 1.65,
|
| 570 |
+
"learning_rate": 1.7754456853171707e-05,
|
| 571 |
+
"loss": 0.2515,
|
| 572 |
+
"step": 18800
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 1.67,
|
| 576 |
+
"learning_rate": 1.7717043605140583e-05,
|
| 577 |
+
"loss": 0.2383,
|
| 578 |
+
"step": 19000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.69,
|
| 582 |
+
"learning_rate": 1.7679630357109454e-05,
|
| 583 |
+
"loss": 0.2522,
|
| 584 |
+
"step": 19200
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.71,
|
| 588 |
+
"learning_rate": 1.7642217109078326e-05,
|
| 589 |
+
"loss": 0.244,
|
| 590 |
+
"step": 19400
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 1.72,
|
| 594 |
+
"learning_rate": 1.7604803861047198e-05,
|
| 595 |
+
"loss": 0.2368,
|
| 596 |
+
"step": 19600
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.74,
|
| 600 |
+
"learning_rate": 1.756739061301607e-05,
|
| 601 |
+
"loss": 0.2416,
|
| 602 |
+
"step": 19800
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.76,
|
| 606 |
+
"learning_rate": 1.7529977364984942e-05,
|
| 607 |
+
"loss": 0.2428,
|
| 608 |
+
"step": 20000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.78,
|
| 612 |
+
"learning_rate": 1.7492564116953814e-05,
|
| 613 |
+
"loss": 0.2387,
|
| 614 |
+
"step": 20200
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.79,
|
| 618 |
+
"learning_rate": 1.745515086892269e-05,
|
| 619 |
+
"loss": 0.2363,
|
| 620 |
+
"step": 20400
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.81,
|
| 624 |
+
"learning_rate": 1.7417737620891558e-05,
|
| 625 |
+
"loss": 0.2433,
|
| 626 |
+
"step": 20600
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.83,
|
| 630 |
+
"learning_rate": 1.738032437286043e-05,
|
| 631 |
+
"loss": 0.2395,
|
| 632 |
+
"step": 20800
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 1.85,
|
| 636 |
+
"learning_rate": 1.7342911124829302e-05,
|
| 637 |
+
"loss": 0.237,
|
| 638 |
+
"step": 21000
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 1.86,
|
| 642 |
+
"learning_rate": 1.7305497876798177e-05,
|
| 643 |
+
"loss": 0.2382,
|
| 644 |
+
"step": 21200
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.88,
|
| 648 |
+
"learning_rate": 1.726808462876705e-05,
|
| 649 |
+
"loss": 0.2306,
|
| 650 |
+
"step": 21400
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.9,
|
| 654 |
+
"learning_rate": 1.723067138073592e-05,
|
| 655 |
+
"loss": 0.2356,
|
| 656 |
+
"step": 21600
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 1.92,
|
| 660 |
+
"learning_rate": 1.7193258132704793e-05,
|
| 661 |
+
"loss": 0.2473,
|
| 662 |
+
"step": 21800
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.93,
|
| 666 |
+
"learning_rate": 1.7155844884673665e-05,
|
| 667 |
+
"loss": 0.2428,
|
| 668 |
+
"step": 22000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.95,
|
| 672 |
+
"learning_rate": 1.7118431636642537e-05,
|
| 673 |
+
"loss": 0.2302,
|
| 674 |
+
"step": 22200
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 1.97,
|
| 678 |
+
"learning_rate": 1.708101838861141e-05,
|
| 679 |
+
"loss": 0.2235,
|
| 680 |
+
"step": 22400
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 1.99,
|
| 684 |
+
"learning_rate": 1.704360514058028e-05,
|
| 685 |
+
"loss": 0.2271,
|
| 686 |
+
"step": 22600
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 2.01,
|
| 690 |
+
"learning_rate": 1.7006191892549152e-05,
|
| 691 |
+
"loss": 0.2229,
|
| 692 |
+
"step": 22800
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 2.02,
|
| 696 |
+
"learning_rate": 1.6968778644518024e-05,
|
| 697 |
+
"loss": 0.2,
|
| 698 |
+
"step": 23000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.04,
|
| 702 |
+
"learning_rate": 1.69313653964869e-05,
|
| 703 |
+
"loss": 0.1905,
|
| 704 |
+
"step": 23200
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.06,
|
| 708 |
+
"learning_rate": 1.689395214845577e-05,
|
| 709 |
+
"loss": 0.1908,
|
| 710 |
+
"step": 23400
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 2.08,
|
| 714 |
+
"learning_rate": 1.685653890042464e-05,
|
| 715 |
+
"loss": 0.2062,
|
| 716 |
+
"step": 23600
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 2.09,
|
| 720 |
+
"learning_rate": 1.6819125652393512e-05,
|
| 721 |
+
"loss": 0.1926,
|
| 722 |
+
"step": 23800
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 2.11,
|
| 726 |
+
"learning_rate": 1.6781712404362387e-05,
|
| 727 |
+
"loss": 0.1867,
|
| 728 |
+
"step": 24000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 2.13,
|
| 732 |
+
"learning_rate": 1.674429915633126e-05,
|
| 733 |
+
"loss": 0.1892,
|
| 734 |
+
"step": 24200
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 2.15,
|
| 738 |
+
"learning_rate": 1.670688590830013e-05,
|
| 739 |
+
"loss": 0.1853,
|
| 740 |
+
"step": 24400
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 2.16,
|
| 744 |
+
"learning_rate": 1.6669472660269003e-05,
|
| 745 |
+
"loss": 0.1972,
|
| 746 |
+
"step": 24600
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 2.18,
|
| 750 |
+
"learning_rate": 1.6632059412237875e-05,
|
| 751 |
+
"loss": 0.1916,
|
| 752 |
+
"step": 24800
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 2.2,
|
| 756 |
+
"learning_rate": 1.6594646164206747e-05,
|
| 757 |
+
"loss": 0.1836,
|
| 758 |
+
"step": 25000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 2.22,
|
| 762 |
+
"learning_rate": 1.655723291617562e-05,
|
| 763 |
+
"loss": 0.1855,
|
| 764 |
+
"step": 25200
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 2.23,
|
| 768 |
+
"learning_rate": 1.651981966814449e-05,
|
| 769 |
+
"loss": 0.2007,
|
| 770 |
+
"step": 25400
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 2.25,
|
| 774 |
+
"learning_rate": 1.6482406420113363e-05,
|
| 775 |
+
"loss": 0.1857,
|
| 776 |
+
"step": 25600
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 2.27,
|
| 780 |
+
"learning_rate": 1.6444993172082235e-05,
|
| 781 |
+
"loss": 0.2067,
|
| 782 |
+
"step": 25800
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 2.29,
|
| 786 |
+
"learning_rate": 1.6407579924051106e-05,
|
| 787 |
+
"loss": 0.2096,
|
| 788 |
+
"step": 26000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.3,
|
| 792 |
+
"learning_rate": 1.6370166676019982e-05,
|
| 793 |
+
"loss": 0.2039,
|
| 794 |
+
"step": 26200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 2.32,
|
| 798 |
+
"learning_rate": 1.6332753427988854e-05,
|
| 799 |
+
"loss": 0.1985,
|
| 800 |
+
"step": 26400
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 2.34,
|
| 804 |
+
"learning_rate": 1.6295340179957722e-05,
|
| 805 |
+
"loss": 0.1927,
|
| 806 |
+
"step": 26600
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.36,
|
| 810 |
+
"learning_rate": 1.6257926931926597e-05,
|
| 811 |
+
"loss": 0.2015,
|
| 812 |
+
"step": 26800
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 2.37,
|
| 816 |
+
"learning_rate": 1.622051368389547e-05,
|
| 817 |
+
"loss": 0.1822,
|
| 818 |
+
"step": 27000
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 2.39,
|
| 822 |
+
"learning_rate": 1.618310043586434e-05,
|
| 823 |
+
"loss": 0.1957,
|
| 824 |
+
"step": 27200
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 2.41,
|
| 828 |
+
"learning_rate": 1.6145687187833213e-05,
|
| 829 |
+
"loss": 0.1936,
|
| 830 |
+
"step": 27400
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 2.43,
|
| 834 |
+
"learning_rate": 1.6108273939802085e-05,
|
| 835 |
+
"loss": 0.1896,
|
| 836 |
+
"step": 27600
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 2.44,
|
| 840 |
+
"learning_rate": 1.6070860691770957e-05,
|
| 841 |
+
"loss": 0.199,
|
| 842 |
+
"step": 27800
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.46,
|
| 846 |
+
"learning_rate": 1.603344744373983e-05,
|
| 847 |
+
"loss": 0.1954,
|
| 848 |
+
"step": 28000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 2.48,
|
| 852 |
+
"learning_rate": 1.5996034195708704e-05,
|
| 853 |
+
"loss": 0.1918,
|
| 854 |
+
"step": 28200
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 2.5,
|
| 858 |
+
"learning_rate": 1.5958620947677573e-05,
|
| 859 |
+
"loss": 0.1902,
|
| 860 |
+
"step": 28400
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.52,
|
| 864 |
+
"learning_rate": 1.5921207699646445e-05,
|
| 865 |
+
"loss": 0.201,
|
| 866 |
+
"step": 28600
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 2.53,
|
| 870 |
+
"learning_rate": 1.5883794451615317e-05,
|
| 871 |
+
"loss": 0.1888,
|
| 872 |
+
"step": 28800
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 2.55,
|
| 876 |
+
"learning_rate": 1.5846381203584192e-05,
|
| 877 |
+
"loss": 0.1932,
|
| 878 |
+
"step": 29000
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.57,
|
| 882 |
+
"learning_rate": 1.5808967955553064e-05,
|
| 883 |
+
"loss": 0.2016,
|
| 884 |
+
"step": 29200
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.59,
|
| 888 |
+
"learning_rate": 1.5771554707521936e-05,
|
| 889 |
+
"loss": 0.199,
|
| 890 |
+
"step": 29400
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 2.6,
|
| 894 |
+
"learning_rate": 1.5734141459490804e-05,
|
| 895 |
+
"loss": 0.1854,
|
| 896 |
+
"step": 29600
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 2.62,
|
| 900 |
+
"learning_rate": 1.569672821145968e-05,
|
| 901 |
+
"loss": 0.1935,
|
| 902 |
+
"step": 29800
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 2.64,
|
| 906 |
+
"learning_rate": 1.565931496342855e-05,
|
| 907 |
+
"loss": 0.1902,
|
| 908 |
+
"step": 30000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 2.66,
|
| 912 |
+
"learning_rate": 1.5621901715397423e-05,
|
| 913 |
+
"loss": 0.1831,
|
| 914 |
+
"step": 30200
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 2.67,
|
| 918 |
+
"learning_rate": 1.5584488467366295e-05,
|
| 919 |
+
"loss": 0.1978,
|
| 920 |
+
"step": 30400
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.69,
|
| 924 |
+
"learning_rate": 1.5547075219335167e-05,
|
| 925 |
+
"loss": 0.1938,
|
| 926 |
+
"step": 30600
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 2.71,
|
| 930 |
+
"learning_rate": 1.550966197130404e-05,
|
| 931 |
+
"loss": 0.1911,
|
| 932 |
+
"step": 30800
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.73,
|
| 936 |
+
"learning_rate": 1.547224872327291e-05,
|
| 937 |
+
"loss": 0.1847,
|
| 938 |
+
"step": 31000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 2.74,
|
| 942 |
+
"learning_rate": 1.5434835475241786e-05,
|
| 943 |
+
"loss": 0.1919,
|
| 944 |
+
"step": 31200
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 2.76,
|
| 948 |
+
"learning_rate": 1.5397422227210655e-05,
|
| 949 |
+
"loss": 0.1944,
|
| 950 |
+
"step": 31400
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 2.78,
|
| 954 |
+
"learning_rate": 1.5360008979179527e-05,
|
| 955 |
+
"loss": 0.1809,
|
| 956 |
+
"step": 31600
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 2.8,
|
| 960 |
+
"learning_rate": 1.5322595731148402e-05,
|
| 961 |
+
"loss": 0.1986,
|
| 962 |
+
"step": 31800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.81,
|
| 966 |
+
"learning_rate": 1.5285182483117274e-05,
|
| 967 |
+
"loss": 0.1927,
|
| 968 |
+
"step": 32000
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 2.83,
|
| 972 |
+
"learning_rate": 1.5247769235086144e-05,
|
| 973 |
+
"loss": 0.1955,
|
| 974 |
+
"step": 32200
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.85,
|
| 978 |
+
"learning_rate": 1.5210355987055016e-05,
|
| 979 |
+
"loss": 0.1909,
|
| 980 |
+
"step": 32400
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 2.87,
|
| 984 |
+
"learning_rate": 1.517294273902389e-05,
|
| 985 |
+
"loss": 0.2026,
|
| 986 |
+
"step": 32600
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 2.88,
|
| 990 |
+
"learning_rate": 1.5135529490992762e-05,
|
| 991 |
+
"loss": 0.1922,
|
| 992 |
+
"step": 32800
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 2.9,
|
| 996 |
+
"learning_rate": 1.5098116242961634e-05,
|
| 997 |
+
"loss": 0.1892,
|
| 998 |
+
"step": 33000
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 2.92,
|
| 1002 |
+
"learning_rate": 1.5060702994930506e-05,
|
| 1003 |
+
"loss": 0.1962,
|
| 1004 |
+
"step": 33200
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 2.94,
|
| 1008 |
+
"learning_rate": 1.502328974689938e-05,
|
| 1009 |
+
"loss": 0.1987,
|
| 1010 |
+
"step": 33400
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 2.95,
|
| 1014 |
+
"learning_rate": 1.4985876498868251e-05,
|
| 1015 |
+
"loss": 0.2,
|
| 1016 |
+
"step": 33600
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.97,
|
| 1020 |
+
"learning_rate": 1.4948463250837121e-05,
|
| 1021 |
+
"loss": 0.1784,
|
| 1022 |
+
"step": 33800
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 2.99,
|
| 1026 |
+
"learning_rate": 1.4911050002805995e-05,
|
| 1027 |
+
"loss": 0.1998,
|
| 1028 |
+
"step": 34000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 3.01,
|
| 1032 |
+
"learning_rate": 1.4873636754774867e-05,
|
| 1033 |
+
"loss": 0.1794,
|
| 1034 |
+
"step": 34200
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 3.03,
|
| 1038 |
+
"learning_rate": 1.4836223506743739e-05,
|
| 1039 |
+
"loss": 0.1545,
|
| 1040 |
+
"step": 34400
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 3.04,
|
| 1044 |
+
"learning_rate": 1.479881025871261e-05,
|
| 1045 |
+
"loss": 0.1596,
|
| 1046 |
+
"step": 34600
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 3.06,
|
| 1050 |
+
"learning_rate": 1.4761397010681484e-05,
|
| 1051 |
+
"loss": 0.1522,
|
| 1052 |
+
"step": 34800
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 3.08,
|
| 1056 |
+
"learning_rate": 1.4723983762650356e-05,
|
| 1057 |
+
"loss": 0.1633,
|
| 1058 |
+
"step": 35000
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 3.1,
|
| 1062 |
+
"learning_rate": 1.4686570514619227e-05,
|
| 1063 |
+
"loss": 0.1421,
|
| 1064 |
+
"step": 35200
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 3.11,
|
| 1068 |
+
"learning_rate": 1.4649157266588102e-05,
|
| 1069 |
+
"loss": 0.1506,
|
| 1070 |
+
"step": 35400
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 3.13,
|
| 1074 |
+
"learning_rate": 1.4611744018556972e-05,
|
| 1075 |
+
"loss": 0.1563,
|
| 1076 |
+
"step": 35600
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 3.15,
|
| 1080 |
+
"learning_rate": 1.4574330770525844e-05,
|
| 1081 |
+
"loss": 0.1557,
|
| 1082 |
+
"step": 35800
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 3.17,
|
| 1086 |
+
"learning_rate": 1.4536917522494716e-05,
|
| 1087 |
+
"loss": 0.1641,
|
| 1088 |
+
"step": 36000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 3.18,
|
| 1092 |
+
"learning_rate": 1.449950427446359e-05,
|
| 1093 |
+
"loss": 0.1545,
|
| 1094 |
+
"step": 36200
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 3.2,
|
| 1098 |
+
"learning_rate": 1.4462091026432461e-05,
|
| 1099 |
+
"loss": 0.1625,
|
| 1100 |
+
"step": 36400
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 3.22,
|
| 1104 |
+
"learning_rate": 1.4424677778401333e-05,
|
| 1105 |
+
"loss": 0.165,
|
| 1106 |
+
"step": 36600
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 3.24,
|
| 1110 |
+
"learning_rate": 1.4387264530370207e-05,
|
| 1111 |
+
"loss": 0.157,
|
| 1112 |
+
"step": 36800
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 3.25,
|
| 1116 |
+
"learning_rate": 1.4349851282339077e-05,
|
| 1117 |
+
"loss": 0.1485,
|
| 1118 |
+
"step": 37000
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 3.27,
|
| 1122 |
+
"learning_rate": 1.4312438034307949e-05,
|
| 1123 |
+
"loss": 0.1571,
|
| 1124 |
+
"step": 37200
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 3.29,
|
| 1128 |
+
"learning_rate": 1.4275024786276821e-05,
|
| 1129 |
+
"loss": 0.1608,
|
| 1130 |
+
"step": 37400
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 3.31,
|
| 1134 |
+
"learning_rate": 1.4237611538245695e-05,
|
| 1135 |
+
"loss": 0.162,
|
| 1136 |
+
"step": 37600
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 3.32,
|
| 1140 |
+
"learning_rate": 1.4200198290214567e-05,
|
| 1141 |
+
"loss": 0.154,
|
| 1142 |
+
"step": 37800
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 3.34,
|
| 1146 |
+
"learning_rate": 1.4162785042183438e-05,
|
| 1147 |
+
"loss": 0.1536,
|
| 1148 |
+
"step": 38000
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 3.36,
|
| 1152 |
+
"learning_rate": 1.4125371794152309e-05,
|
| 1153 |
+
"loss": 0.164,
|
| 1154 |
+
"step": 38200
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 3.38,
|
| 1158 |
+
"learning_rate": 1.4087958546121184e-05,
|
| 1159 |
+
"loss": 0.1655,
|
| 1160 |
+
"step": 38400
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 3.39,
|
| 1164 |
+
"learning_rate": 1.4050545298090054e-05,
|
| 1165 |
+
"loss": 0.1606,
|
| 1166 |
+
"step": 38600
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 3.41,
|
| 1170 |
+
"learning_rate": 1.4013132050058926e-05,
|
| 1171 |
+
"loss": 0.1533,
|
| 1172 |
+
"step": 38800
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 3.43,
|
| 1176 |
+
"learning_rate": 1.39757188020278e-05,
|
| 1177 |
+
"loss": 0.1515,
|
| 1178 |
+
"step": 39000
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 3.45,
|
| 1182 |
+
"learning_rate": 1.3938305553996672e-05,
|
| 1183 |
+
"loss": 0.1624,
|
| 1184 |
+
"step": 39200
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 3.46,
|
| 1188 |
+
"learning_rate": 1.3900892305965544e-05,
|
| 1189 |
+
"loss": 0.1467,
|
| 1190 |
+
"step": 39400
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 3.48,
|
| 1194 |
+
"learning_rate": 1.3863479057934415e-05,
|
| 1195 |
+
"loss": 0.1582,
|
| 1196 |
+
"step": 39600
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 3.5,
|
| 1200 |
+
"learning_rate": 1.3826065809903289e-05,
|
| 1201 |
+
"loss": 0.163,
|
| 1202 |
+
"step": 39800
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 3.52,
|
| 1206 |
+
"learning_rate": 1.378865256187216e-05,
|
| 1207 |
+
"loss": 0.1561,
|
| 1208 |
+
"step": 40000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 3.54,
|
| 1212 |
+
"learning_rate": 1.3751239313841031e-05,
|
| 1213 |
+
"loss": 0.1558,
|
| 1214 |
+
"step": 40200
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 3.55,
|
| 1218 |
+
"learning_rate": 1.3713826065809905e-05,
|
| 1219 |
+
"loss": 0.1686,
|
| 1220 |
+
"step": 40400
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 3.57,
|
| 1224 |
+
"learning_rate": 1.3676412817778777e-05,
|
| 1225 |
+
"loss": 0.161,
|
| 1226 |
+
"step": 40600
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 3.59,
|
| 1230 |
+
"learning_rate": 1.3638999569747649e-05,
|
| 1231 |
+
"loss": 0.1574,
|
| 1232 |
+
"step": 40800
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 3.61,
|
| 1236 |
+
"learning_rate": 1.360158632171652e-05,
|
| 1237 |
+
"loss": 0.1591,
|
| 1238 |
+
"step": 41000
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 3.62,
|
| 1242 |
+
"learning_rate": 1.3564173073685394e-05,
|
| 1243 |
+
"loss": 0.1618,
|
| 1244 |
+
"step": 41200
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 3.64,
|
| 1248 |
+
"learning_rate": 1.3526759825654266e-05,
|
| 1249 |
+
"loss": 0.1545,
|
| 1250 |
+
"step": 41400
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 3.66,
|
| 1254 |
+
"learning_rate": 1.3489346577623136e-05,
|
| 1255 |
+
"loss": 0.1626,
|
| 1256 |
+
"step": 41600
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 3.68,
|
| 1260 |
+
"learning_rate": 1.345193332959201e-05,
|
| 1261 |
+
"loss": 0.1588,
|
| 1262 |
+
"step": 41800
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 3.69,
|
| 1266 |
+
"learning_rate": 1.3414520081560882e-05,
|
| 1267 |
+
"loss": 0.1538,
|
| 1268 |
+
"step": 42000
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 3.71,
|
| 1272 |
+
"learning_rate": 1.3377106833529754e-05,
|
| 1273 |
+
"loss": 0.1557,
|
| 1274 |
+
"step": 42200
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 3.73,
|
| 1278 |
+
"learning_rate": 1.3339693585498626e-05,
|
| 1279 |
+
"loss": 0.1554,
|
| 1280 |
+
"step": 42400
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 3.75,
|
| 1284 |
+
"learning_rate": 1.33022803374675e-05,
|
| 1285 |
+
"loss": 0.1639,
|
| 1286 |
+
"step": 42600
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 3.76,
|
| 1290 |
+
"learning_rate": 1.3264867089436371e-05,
|
| 1291 |
+
"loss": 0.1519,
|
| 1292 |
+
"step": 42800
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 3.78,
|
| 1296 |
+
"learning_rate": 1.3227453841405241e-05,
|
| 1297 |
+
"loss": 0.1671,
|
| 1298 |
+
"step": 43000
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 3.8,
|
| 1302 |
+
"learning_rate": 1.3190040593374113e-05,
|
| 1303 |
+
"loss": 0.1614,
|
| 1304 |
+
"step": 43200
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 3.82,
|
| 1308 |
+
"learning_rate": 1.3152627345342987e-05,
|
| 1309 |
+
"loss": 0.1523,
|
| 1310 |
+
"step": 43400
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 3.83,
|
| 1314 |
+
"learning_rate": 1.3115214097311859e-05,
|
| 1315 |
+
"loss": 0.1564,
|
| 1316 |
+
"step": 43600
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 3.85,
|
| 1320 |
+
"learning_rate": 1.307780084928073e-05,
|
| 1321 |
+
"loss": 0.1662,
|
| 1322 |
+
"step": 43800
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 3.87,
|
| 1326 |
+
"learning_rate": 1.3040387601249604e-05,
|
| 1327 |
+
"loss": 0.1667,
|
| 1328 |
+
"step": 44000
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 3.89,
|
| 1332 |
+
"learning_rate": 1.3002974353218476e-05,
|
| 1333 |
+
"loss": 0.1631,
|
| 1334 |
+
"step": 44200
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 3.9,
|
| 1338 |
+
"learning_rate": 1.2965561105187348e-05,
|
| 1339 |
+
"loss": 0.1561,
|
| 1340 |
+
"step": 44400
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 3.92,
|
| 1344 |
+
"learning_rate": 1.2928147857156219e-05,
|
| 1345 |
+
"loss": 0.159,
|
| 1346 |
+
"step": 44600
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 3.94,
|
| 1350 |
+
"learning_rate": 1.2890734609125092e-05,
|
| 1351 |
+
"loss": 0.1618,
|
| 1352 |
+
"step": 44800
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 3.96,
|
| 1356 |
+
"learning_rate": 1.2853321361093964e-05,
|
| 1357 |
+
"loss": 0.1538,
|
| 1358 |
+
"step": 45000
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 3.98,
|
| 1362 |
+
"learning_rate": 1.2815908113062836e-05,
|
| 1363 |
+
"loss": 0.1586,
|
| 1364 |
+
"step": 45200
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 3.99,
|
| 1368 |
+
"learning_rate": 1.277849486503171e-05,
|
| 1369 |
+
"loss": 0.16,
|
| 1370 |
+
"step": 45400
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 4.01,
|
| 1374 |
+
"learning_rate": 1.2741081617000581e-05,
|
| 1375 |
+
"loss": 0.1483,
|
| 1376 |
+
"step": 45600
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 4.03,
|
| 1380 |
+
"learning_rate": 1.2703668368969453e-05,
|
| 1381 |
+
"loss": 0.1201,
|
| 1382 |
+
"step": 45800
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 4.05,
|
| 1386 |
+
"learning_rate": 1.2666255120938324e-05,
|
| 1387 |
+
"loss": 0.1278,
|
| 1388 |
+
"step": 46000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 4.06,
|
| 1392 |
+
"learning_rate": 1.2628841872907199e-05,
|
| 1393 |
+
"loss": 0.1348,
|
| 1394 |
+
"step": 46200
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 4.08,
|
| 1398 |
+
"learning_rate": 1.2591428624876069e-05,
|
| 1399 |
+
"loss": 0.1238,
|
| 1400 |
+
"step": 46400
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 4.1,
|
| 1404 |
+
"learning_rate": 1.2554015376844941e-05,
|
| 1405 |
+
"loss": 0.1215,
|
| 1406 |
+
"step": 46600
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 4.12,
|
| 1410 |
+
"learning_rate": 1.2516602128813815e-05,
|
| 1411 |
+
"loss": 0.132,
|
| 1412 |
+
"step": 46800
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 4.13,
|
| 1416 |
+
"learning_rate": 1.2479188880782687e-05,
|
| 1417 |
+
"loss": 0.1244,
|
| 1418 |
+
"step": 47000
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 4.15,
|
| 1422 |
+
"learning_rate": 1.2441775632751559e-05,
|
| 1423 |
+
"loss": 0.1235,
|
| 1424 |
+
"step": 47200
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 4.17,
|
| 1428 |
+
"learning_rate": 1.240436238472043e-05,
|
| 1429 |
+
"loss": 0.1312,
|
| 1430 |
+
"step": 47400
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 4.19,
|
| 1434 |
+
"learning_rate": 1.2366949136689304e-05,
|
| 1435 |
+
"loss": 0.1226,
|
| 1436 |
+
"step": 47600
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 4.2,
|
| 1440 |
+
"learning_rate": 1.2329535888658174e-05,
|
| 1441 |
+
"loss": 0.1292,
|
| 1442 |
+
"step": 47800
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 4.22,
|
| 1446 |
+
"learning_rate": 1.2292122640627046e-05,
|
| 1447 |
+
"loss": 0.1255,
|
| 1448 |
+
"step": 48000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 4.24,
|
| 1452 |
+
"learning_rate": 1.2254709392595918e-05,
|
| 1453 |
+
"loss": 0.1258,
|
| 1454 |
+
"step": 48200
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 4.26,
|
| 1458 |
+
"learning_rate": 1.2217296144564792e-05,
|
| 1459 |
+
"loss": 0.1292,
|
| 1460 |
+
"step": 48400
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 4.27,
|
| 1464 |
+
"learning_rate": 1.2179882896533664e-05,
|
| 1465 |
+
"loss": 0.1298,
|
| 1466 |
+
"step": 48600
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 4.29,
|
| 1470 |
+
"learning_rate": 1.2142469648502536e-05,
|
| 1471 |
+
"loss": 0.1254,
|
| 1472 |
+
"step": 48800
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 4.31,
|
| 1476 |
+
"learning_rate": 1.210505640047141e-05,
|
| 1477 |
+
"loss": 0.1241,
|
| 1478 |
+
"step": 49000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 4.33,
|
| 1482 |
+
"learning_rate": 1.2067643152440281e-05,
|
| 1483 |
+
"loss": 0.1302,
|
| 1484 |
+
"step": 49200
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 4.34,
|
| 1488 |
+
"learning_rate": 1.2030229904409151e-05,
|
| 1489 |
+
"loss": 0.1309,
|
| 1490 |
+
"step": 49400
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 4.36,
|
| 1494 |
+
"learning_rate": 1.1992816656378023e-05,
|
| 1495 |
+
"loss": 0.1182,
|
| 1496 |
+
"step": 49600
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 4.38,
|
| 1500 |
+
"learning_rate": 1.1955403408346897e-05,
|
| 1501 |
+
"loss": 0.1331,
|
| 1502 |
+
"step": 49800
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 4.4,
|
| 1506 |
+
"learning_rate": 1.1917990160315769e-05,
|
| 1507 |
+
"loss": 0.1289,
|
| 1508 |
+
"step": 50000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 4.41,
|
| 1512 |
+
"learning_rate": 1.188057691228464e-05,
|
| 1513 |
+
"loss": 0.1149,
|
| 1514 |
+
"step": 50200
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 4.43,
|
| 1518 |
+
"learning_rate": 1.1843163664253514e-05,
|
| 1519 |
+
"loss": 0.1201,
|
| 1520 |
+
"step": 50400
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 4.45,
|
| 1524 |
+
"learning_rate": 1.1805750416222386e-05,
|
| 1525 |
+
"loss": 0.1218,
|
| 1526 |
+
"step": 50600
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 4.47,
|
| 1530 |
+
"learning_rate": 1.1768337168191256e-05,
|
| 1531 |
+
"loss": 0.1278,
|
| 1532 |
+
"step": 50800
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 4.49,
|
| 1536 |
+
"learning_rate": 1.1730923920160128e-05,
|
| 1537 |
+
"loss": 0.1275,
|
| 1538 |
+
"step": 51000
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 4.5,
|
| 1542 |
+
"learning_rate": 1.1693510672129002e-05,
|
| 1543 |
+
"loss": 0.1379,
|
| 1544 |
+
"step": 51200
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 4.52,
|
| 1548 |
+
"learning_rate": 1.1656097424097874e-05,
|
| 1549 |
+
"loss": 0.1298,
|
| 1550 |
+
"step": 51400
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 4.54,
|
| 1554 |
+
"learning_rate": 1.1618684176066746e-05,
|
| 1555 |
+
"loss": 0.1375,
|
| 1556 |
+
"step": 51600
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 4.56,
|
| 1560 |
+
"learning_rate": 1.1581270928035618e-05,
|
| 1561 |
+
"loss": 0.1271,
|
| 1562 |
+
"step": 51800
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 4.57,
|
| 1566 |
+
"learning_rate": 1.1543857680004491e-05,
|
| 1567 |
+
"loss": 0.1487,
|
| 1568 |
+
"step": 52000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 4.59,
|
| 1572 |
+
"learning_rate": 1.1506444431973363e-05,
|
| 1573 |
+
"loss": 0.1341,
|
| 1574 |
+
"step": 52200
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 4.61,
|
| 1578 |
+
"learning_rate": 1.1469031183942233e-05,
|
| 1579 |
+
"loss": 0.1314,
|
| 1580 |
+
"step": 52400
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 4.63,
|
| 1584 |
+
"learning_rate": 1.1431617935911107e-05,
|
| 1585 |
+
"loss": 0.1187,
|
| 1586 |
+
"step": 52600
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 4.64,
|
| 1590 |
+
"learning_rate": 1.1394204687879979e-05,
|
| 1591 |
+
"loss": 0.1308,
|
| 1592 |
+
"step": 52800
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 4.66,
|
| 1596 |
+
"learning_rate": 1.1356791439848851e-05,
|
| 1597 |
+
"loss": 0.1357,
|
| 1598 |
+
"step": 53000
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 4.68,
|
| 1602 |
+
"learning_rate": 1.1319378191817723e-05,
|
| 1603 |
+
"loss": 0.1346,
|
| 1604 |
+
"step": 53200
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 4.7,
|
| 1608 |
+
"learning_rate": 1.1281964943786596e-05,
|
| 1609 |
+
"loss": 0.1302,
|
| 1610 |
+
"step": 53400
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 4.71,
|
| 1614 |
+
"learning_rate": 1.1244551695755468e-05,
|
| 1615 |
+
"loss": 0.1318,
|
| 1616 |
+
"step": 53600
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 4.73,
|
| 1620 |
+
"learning_rate": 1.120713844772434e-05,
|
| 1621 |
+
"loss": 0.1406,
|
| 1622 |
+
"step": 53800
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 4.75,
|
| 1626 |
+
"learning_rate": 1.1169725199693214e-05,
|
| 1627 |
+
"loss": 0.1287,
|
| 1628 |
+
"step": 54000
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 4.77,
|
| 1632 |
+
"learning_rate": 1.1132311951662084e-05,
|
| 1633 |
+
"loss": 0.1296,
|
| 1634 |
+
"step": 54200
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 4.78,
|
| 1638 |
+
"learning_rate": 1.1094898703630956e-05,
|
| 1639 |
+
"loss": 0.1239,
|
| 1640 |
+
"step": 54400
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 4.8,
|
| 1644 |
+
"learning_rate": 1.1057485455599828e-05,
|
| 1645 |
+
"loss": 0.1289,
|
| 1646 |
+
"step": 54600
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 4.82,
|
| 1650 |
+
"learning_rate": 1.1020072207568702e-05,
|
| 1651 |
+
"loss": 0.1371,
|
| 1652 |
+
"step": 54800
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 4.84,
|
| 1656 |
+
"learning_rate": 1.0982658959537573e-05,
|
| 1657 |
+
"loss": 0.1371,
|
| 1658 |
+
"step": 55000
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 4.85,
|
| 1662 |
+
"learning_rate": 1.0945245711506445e-05,
|
| 1663 |
+
"loss": 0.1197,
|
| 1664 |
+
"step": 55200
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 4.87,
|
| 1668 |
+
"learning_rate": 1.0907832463475319e-05,
|
| 1669 |
+
"loss": 0.1316,
|
| 1670 |
+
"step": 55400
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 4.89,
|
| 1674 |
+
"learning_rate": 1.0870419215444191e-05,
|
| 1675 |
+
"loss": 0.1275,
|
| 1676 |
+
"step": 55600
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 4.91,
|
| 1680 |
+
"learning_rate": 1.0833005967413061e-05,
|
| 1681 |
+
"loss": 0.1287,
|
| 1682 |
+
"step": 55800
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 4.92,
|
| 1686 |
+
"learning_rate": 1.0795592719381933e-05,
|
| 1687 |
+
"loss": 0.1266,
|
| 1688 |
+
"step": 56000
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 4.94,
|
| 1692 |
+
"learning_rate": 1.0758179471350807e-05,
|
| 1693 |
+
"loss": 0.1275,
|
| 1694 |
+
"step": 56200
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 4.96,
|
| 1698 |
+
"learning_rate": 1.0720766223319679e-05,
|
| 1699 |
+
"loss": 0.1244,
|
| 1700 |
+
"step": 56400
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 4.98,
|
| 1704 |
+
"learning_rate": 1.068335297528855e-05,
|
| 1705 |
+
"loss": 0.1453,
|
| 1706 |
+
"step": 56600
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 5.0,
|
| 1710 |
+
"learning_rate": 1.0645939727257422e-05,
|
| 1711 |
+
"loss": 0.1343,
|
| 1712 |
+
"step": 56800
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 5.01,
|
| 1716 |
+
"learning_rate": 1.0608526479226296e-05,
|
| 1717 |
+
"loss": 0.114,
|
| 1718 |
+
"step": 57000
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 5.03,
|
| 1722 |
+
"learning_rate": 1.0571113231195166e-05,
|
| 1723 |
+
"loss": 0.1038,
|
| 1724 |
+
"step": 57200
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 5.05,
|
| 1728 |
+
"learning_rate": 1.0533699983164038e-05,
|
| 1729 |
+
"loss": 0.1064,
|
| 1730 |
+
"step": 57400
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 5.07,
|
| 1734 |
+
"learning_rate": 1.0496286735132912e-05,
|
| 1735 |
+
"loss": 0.0928,
|
| 1736 |
+
"step": 57600
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 5.08,
|
| 1740 |
+
"learning_rate": 1.0458873487101784e-05,
|
| 1741 |
+
"loss": 0.1079,
|
| 1742 |
+
"step": 57800
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 5.1,
|
| 1746 |
+
"learning_rate": 1.0421460239070656e-05,
|
| 1747 |
+
"loss": 0.0976,
|
| 1748 |
+
"step": 58000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 5.12,
|
| 1752 |
+
"learning_rate": 1.0384046991039528e-05,
|
| 1753 |
+
"loss": 0.1086,
|
| 1754 |
+
"step": 58200
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 5.14,
|
| 1758 |
+
"learning_rate": 1.0346633743008401e-05,
|
| 1759 |
+
"loss": 0.105,
|
| 1760 |
+
"step": 58400
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 5.15,
|
| 1764 |
+
"learning_rate": 1.0309220494977273e-05,
|
| 1765 |
+
"loss": 0.1086,
|
| 1766 |
+
"step": 58600
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 5.17,
|
| 1770 |
+
"learning_rate": 1.0271807246946143e-05,
|
| 1771 |
+
"loss": 0.0972,
|
| 1772 |
+
"step": 58800
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 5.19,
|
| 1776 |
+
"learning_rate": 1.0234393998915017e-05,
|
| 1777 |
+
"loss": 0.1086,
|
| 1778 |
+
"step": 59000
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 5.21,
|
| 1782 |
+
"learning_rate": 1.0196980750883889e-05,
|
| 1783 |
+
"loss": 0.1151,
|
| 1784 |
+
"step": 59200
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 5.22,
|
| 1788 |
+
"learning_rate": 1.015956750285276e-05,
|
| 1789 |
+
"loss": 0.1076,
|
| 1790 |
+
"step": 59400
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 5.24,
|
| 1794 |
+
"learning_rate": 1.0122154254821633e-05,
|
| 1795 |
+
"loss": 0.1061,
|
| 1796 |
+
"step": 59600
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 5.26,
|
| 1800 |
+
"learning_rate": 1.0084741006790506e-05,
|
| 1801 |
+
"loss": 0.1087,
|
| 1802 |
+
"step": 59800
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 5.28,
|
| 1806 |
+
"learning_rate": 1.0047327758759378e-05,
|
| 1807 |
+
"loss": 0.1081,
|
| 1808 |
+
"step": 60000
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 5.29,
|
| 1812 |
+
"learning_rate": 1.0009914510728248e-05,
|
| 1813 |
+
"loss": 0.1051,
|
| 1814 |
+
"step": 60200
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 5.31,
|
| 1818 |
+
"learning_rate": 9.972501262697122e-06,
|
| 1819 |
+
"loss": 0.1096,
|
| 1820 |
+
"step": 60400
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 5.33,
|
| 1824 |
+
"learning_rate": 9.935088014665994e-06,
|
| 1825 |
+
"loss": 0.1067,
|
| 1826 |
+
"step": 60600
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 5.35,
|
| 1830 |
+
"learning_rate": 9.897674766634866e-06,
|
| 1831 |
+
"loss": 0.1106,
|
| 1832 |
+
"step": 60800
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 5.36,
|
| 1836 |
+
"learning_rate": 9.86026151860374e-06,
|
| 1837 |
+
"loss": 0.1019,
|
| 1838 |
+
"step": 61000
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 5.38,
|
| 1842 |
+
"learning_rate": 9.82284827057261e-06,
|
| 1843 |
+
"loss": 0.1107,
|
| 1844 |
+
"step": 61200
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 5.4,
|
| 1848 |
+
"learning_rate": 9.785435022541483e-06,
|
| 1849 |
+
"loss": 0.1045,
|
| 1850 |
+
"step": 61400
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 5.42,
|
| 1854 |
+
"learning_rate": 9.748021774510355e-06,
|
| 1855 |
+
"loss": 0.1101,
|
| 1856 |
+
"step": 61600
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 5.43,
|
| 1860 |
+
"learning_rate": 9.710608526479227e-06,
|
| 1861 |
+
"loss": 0.1001,
|
| 1862 |
+
"step": 61800
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 5.45,
|
| 1866 |
+
"learning_rate": 9.673195278448099e-06,
|
| 1867 |
+
"loss": 0.1093,
|
| 1868 |
+
"step": 62000
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 5.47,
|
| 1872 |
+
"learning_rate": 9.635782030416971e-06,
|
| 1873 |
+
"loss": 0.1091,
|
| 1874 |
+
"step": 62200
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 5.49,
|
| 1878 |
+
"learning_rate": 9.598368782385845e-06,
|
| 1879 |
+
"loss": 0.1073,
|
| 1880 |
+
"step": 62400
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 5.51,
|
| 1884 |
+
"learning_rate": 9.560955534354715e-06,
|
| 1885 |
+
"loss": 0.1175,
|
| 1886 |
+
"step": 62600
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 5.52,
|
| 1890 |
+
"learning_rate": 9.523542286323588e-06,
|
| 1891 |
+
"loss": 0.1018,
|
| 1892 |
+
"step": 62800
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 5.54,
|
| 1896 |
+
"learning_rate": 9.48612903829246e-06,
|
| 1897 |
+
"loss": 0.1015,
|
| 1898 |
+
"step": 63000
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 5.56,
|
| 1902 |
+
"learning_rate": 9.448715790261332e-06,
|
| 1903 |
+
"loss": 0.1159,
|
| 1904 |
+
"step": 63200
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 5.58,
|
| 1908 |
+
"learning_rate": 9.411302542230204e-06,
|
| 1909 |
+
"loss": 0.1104,
|
| 1910 |
+
"step": 63400
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 5.59,
|
| 1914 |
+
"learning_rate": 9.373889294199076e-06,
|
| 1915 |
+
"loss": 0.1105,
|
| 1916 |
+
"step": 63600
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 5.61,
|
| 1920 |
+
"learning_rate": 9.33647604616795e-06,
|
| 1921 |
+
"loss": 0.1037,
|
| 1922 |
+
"step": 63800
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 5.63,
|
| 1926 |
+
"learning_rate": 9.299062798136822e-06,
|
| 1927 |
+
"loss": 0.103,
|
| 1928 |
+
"step": 64000
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 5.65,
|
| 1932 |
+
"learning_rate": 9.261649550105694e-06,
|
| 1933 |
+
"loss": 0.1129,
|
| 1934 |
+
"step": 64200
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 5.66,
|
| 1938 |
+
"learning_rate": 9.224236302074565e-06,
|
| 1939 |
+
"loss": 0.1005,
|
| 1940 |
+
"step": 64400
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 5.68,
|
| 1944 |
+
"learning_rate": 9.186823054043437e-06,
|
| 1945 |
+
"loss": 0.1082,
|
| 1946 |
+
"step": 64600
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 5.7,
|
| 1950 |
+
"learning_rate": 9.14940980601231e-06,
|
| 1951 |
+
"loss": 0.1157,
|
| 1952 |
+
"step": 64800
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 5.72,
|
| 1956 |
+
"learning_rate": 9.111996557981181e-06,
|
| 1957 |
+
"loss": 0.1139,
|
| 1958 |
+
"step": 65000
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 5.73,
|
| 1962 |
+
"learning_rate": 9.074583309950053e-06,
|
| 1963 |
+
"loss": 0.1101,
|
| 1964 |
+
"step": 65200
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 5.75,
|
| 1968 |
+
"learning_rate": 9.037170061918927e-06,
|
| 1969 |
+
"loss": 0.1139,
|
| 1970 |
+
"step": 65400
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 5.77,
|
| 1974 |
+
"learning_rate": 8.999756813887799e-06,
|
| 1975 |
+
"loss": 0.1107,
|
| 1976 |
+
"step": 65600
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 5.79,
|
| 1980 |
+
"learning_rate": 8.96234356585667e-06,
|
| 1981 |
+
"loss": 0.1095,
|
| 1982 |
+
"step": 65800
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 5.8,
|
| 1986 |
+
"learning_rate": 8.924930317825543e-06,
|
| 1987 |
+
"loss": 0.1127,
|
| 1988 |
+
"step": 66000
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 5.82,
|
| 1992 |
+
"learning_rate": 8.887517069794414e-06,
|
| 1993 |
+
"loss": 0.1118,
|
| 1994 |
+
"step": 66200
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 5.84,
|
| 1998 |
+
"learning_rate": 8.850103821763288e-06,
|
| 1999 |
+
"loss": 0.1042,
|
| 2000 |
+
"step": 66400
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 5.86,
|
| 2004 |
+
"learning_rate": 8.812690573732158e-06,
|
| 2005 |
+
"loss": 0.1112,
|
| 2006 |
+
"step": 66600
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 5.87,
|
| 2010 |
+
"learning_rate": 8.775277325701032e-06,
|
| 2011 |
+
"loss": 0.1116,
|
| 2012 |
+
"step": 66800
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 5.89,
|
| 2016 |
+
"learning_rate": 8.737864077669904e-06,
|
| 2017 |
+
"loss": 0.1139,
|
| 2018 |
+
"step": 67000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 5.91,
|
| 2022 |
+
"learning_rate": 8.700450829638776e-06,
|
| 2023 |
+
"loss": 0.1082,
|
| 2024 |
+
"step": 67200
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 5.93,
|
| 2028 |
+
"learning_rate": 8.663037581607648e-06,
|
| 2029 |
+
"loss": 0.1056,
|
| 2030 |
+
"step": 67400
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 5.94,
|
| 2034 |
+
"learning_rate": 8.62562433357652e-06,
|
| 2035 |
+
"loss": 0.102,
|
| 2036 |
+
"step": 67600
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 5.96,
|
| 2040 |
+
"learning_rate": 8.588211085545393e-06,
|
| 2041 |
+
"loss": 0.1026,
|
| 2042 |
+
"step": 67800
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 5.98,
|
| 2046 |
+
"learning_rate": 8.550797837514263e-06,
|
| 2047 |
+
"loss": 0.1103,
|
| 2048 |
+
"step": 68000
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 6.0,
|
| 2052 |
+
"learning_rate": 8.513384589483137e-06,
|
| 2053 |
+
"loss": 0.1147,
|
| 2054 |
+
"step": 68200
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 6.02,
|
| 2058 |
+
"learning_rate": 8.475971341452009e-06,
|
| 2059 |
+
"loss": 0.0773,
|
| 2060 |
+
"step": 68400
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 6.03,
|
| 2064 |
+
"learning_rate": 8.43855809342088e-06,
|
| 2065 |
+
"loss": 0.0812,
|
| 2066 |
+
"step": 68600
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 6.05,
|
| 2070 |
+
"learning_rate": 8.401144845389754e-06,
|
| 2071 |
+
"loss": 0.0801,
|
| 2072 |
+
"step": 68800
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 6.07,
|
| 2076 |
+
"learning_rate": 8.363731597358625e-06,
|
| 2077 |
+
"loss": 0.0884,
|
| 2078 |
+
"step": 69000
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 6.09,
|
| 2082 |
+
"learning_rate": 8.326318349327498e-06,
|
| 2083 |
+
"loss": 0.0914,
|
| 2084 |
+
"step": 69200
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 6.1,
|
| 2088 |
+
"learning_rate": 8.28890510129637e-06,
|
| 2089 |
+
"loss": 0.0868,
|
| 2090 |
+
"step": 69400
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 6.12,
|
| 2094 |
+
"learning_rate": 8.251491853265242e-06,
|
| 2095 |
+
"loss": 0.0948,
|
| 2096 |
+
"step": 69600
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 6.14,
|
| 2100 |
+
"learning_rate": 8.214078605234114e-06,
|
| 2101 |
+
"loss": 0.0808,
|
| 2102 |
+
"step": 69800
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 6.16,
|
| 2106 |
+
"learning_rate": 8.176665357202986e-06,
|
| 2107 |
+
"loss": 0.092,
|
| 2108 |
+
"step": 70000
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 6.17,
|
| 2112 |
+
"learning_rate": 8.139252109171858e-06,
|
| 2113 |
+
"loss": 0.0841,
|
| 2114 |
+
"step": 70200
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 6.19,
|
| 2118 |
+
"learning_rate": 8.10183886114073e-06,
|
| 2119 |
+
"loss": 0.0951,
|
| 2120 |
+
"step": 70400
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 6.21,
|
| 2124 |
+
"learning_rate": 8.064425613109603e-06,
|
| 2125 |
+
"loss": 0.0928,
|
| 2126 |
+
"step": 70600
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 6.23,
|
| 2130 |
+
"learning_rate": 8.027012365078475e-06,
|
| 2131 |
+
"loss": 0.0935,
|
| 2132 |
+
"step": 70800
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 6.24,
|
| 2136 |
+
"learning_rate": 7.989599117047347e-06,
|
| 2137 |
+
"loss": 0.0927,
|
| 2138 |
+
"step": 71000
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 6.26,
|
| 2142 |
+
"learning_rate": 7.952185869016219e-06,
|
| 2143 |
+
"loss": 0.0923,
|
| 2144 |
+
"step": 71200
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 6.28,
|
| 2148 |
+
"learning_rate": 7.914772620985091e-06,
|
| 2149 |
+
"loss": 0.0801,
|
| 2150 |
+
"step": 71400
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 6.3,
|
| 2154 |
+
"learning_rate": 7.877359372953963e-06,
|
| 2155 |
+
"loss": 0.0937,
|
| 2156 |
+
"step": 71600
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 6.31,
|
| 2160 |
+
"learning_rate": 7.839946124922837e-06,
|
| 2161 |
+
"loss": 0.0865,
|
| 2162 |
+
"step": 71800
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 6.33,
|
| 2166 |
+
"learning_rate": 7.802532876891707e-06,
|
| 2167 |
+
"loss": 0.0871,
|
| 2168 |
+
"step": 72000
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 6.35,
|
| 2172 |
+
"learning_rate": 7.76511962886058e-06,
|
| 2173 |
+
"loss": 0.0786,
|
| 2174 |
+
"step": 72200
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 6.37,
|
| 2178 |
+
"learning_rate": 7.727706380829452e-06,
|
| 2179 |
+
"loss": 0.0934,
|
| 2180 |
+
"step": 72400
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 6.38,
|
| 2184 |
+
"learning_rate": 7.690293132798324e-06,
|
| 2185 |
+
"loss": 0.0838,
|
| 2186 |
+
"step": 72600
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 6.4,
|
| 2190 |
+
"learning_rate": 7.652879884767196e-06,
|
| 2191 |
+
"loss": 0.097,
|
| 2192 |
+
"step": 72800
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 6.42,
|
| 2196 |
+
"learning_rate": 7.615466636736069e-06,
|
| 2197 |
+
"loss": 0.0885,
|
| 2198 |
+
"step": 73000
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 6.44,
|
| 2202 |
+
"learning_rate": 7.578053388704941e-06,
|
| 2203 |
+
"loss": 0.0919,
|
| 2204 |
+
"step": 73200
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 6.46,
|
| 2208 |
+
"learning_rate": 7.540640140673813e-06,
|
| 2209 |
+
"loss": 0.0822,
|
| 2210 |
+
"step": 73400
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 6.47,
|
| 2214 |
+
"learning_rate": 7.5032268926426856e-06,
|
| 2215 |
+
"loss": 0.0837,
|
| 2216 |
+
"step": 73600
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 6.49,
|
| 2220 |
+
"learning_rate": 7.465813644611558e-06,
|
| 2221 |
+
"loss": 0.0879,
|
| 2222 |
+
"step": 73800
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 6.51,
|
| 2226 |
+
"learning_rate": 7.428400396580429e-06,
|
| 2227 |
+
"loss": 0.0927,
|
| 2228 |
+
"step": 74000
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 6.53,
|
| 2232 |
+
"learning_rate": 7.390987148549302e-06,
|
| 2233 |
+
"loss": 0.0929,
|
| 2234 |
+
"step": 74200
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 6.54,
|
| 2238 |
+
"learning_rate": 7.353573900518174e-06,
|
| 2239 |
+
"loss": 0.0871,
|
| 2240 |
+
"step": 74400
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 6.56,
|
| 2244 |
+
"learning_rate": 7.316160652487047e-06,
|
| 2245 |
+
"loss": 0.0886,
|
| 2246 |
+
"step": 74600
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 6.58,
|
| 2250 |
+
"learning_rate": 7.278747404455918e-06,
|
| 2251 |
+
"loss": 0.0887,
|
| 2252 |
+
"step": 74800
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 6.6,
|
| 2256 |
+
"learning_rate": 7.241334156424791e-06,
|
| 2257 |
+
"loss": 0.0924,
|
| 2258 |
+
"step": 75000
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 6.61,
|
| 2262 |
+
"learning_rate": 7.203920908393663e-06,
|
| 2263 |
+
"loss": 0.0971,
|
| 2264 |
+
"step": 75200
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 6.63,
|
| 2268 |
+
"learning_rate": 7.166507660362535e-06,
|
| 2269 |
+
"loss": 0.0922,
|
| 2270 |
+
"step": 75400
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 6.65,
|
| 2274 |
+
"learning_rate": 7.129094412331407e-06,
|
| 2275 |
+
"loss": 0.0866,
|
| 2276 |
+
"step": 75600
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 6.67,
|
| 2280 |
+
"learning_rate": 7.091681164300279e-06,
|
| 2281 |
+
"loss": 0.0822,
|
| 2282 |
+
"step": 75800
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 6.68,
|
| 2286 |
+
"learning_rate": 7.054267916269152e-06,
|
| 2287 |
+
"loss": 0.102,
|
| 2288 |
+
"step": 76000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 6.7,
|
| 2292 |
+
"learning_rate": 7.016854668238023e-06,
|
| 2293 |
+
"loss": 0.091,
|
| 2294 |
+
"step": 76200
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 6.72,
|
| 2298 |
+
"learning_rate": 6.979441420206896e-06,
|
| 2299 |
+
"loss": 0.0937,
|
| 2300 |
+
"step": 76400
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 6.74,
|
| 2304 |
+
"learning_rate": 6.942028172175768e-06,
|
| 2305 |
+
"loss": 0.0795,
|
| 2306 |
+
"step": 76600
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 6.75,
|
| 2310 |
+
"learning_rate": 6.9046149241446405e-06,
|
| 2311 |
+
"loss": 0.0917,
|
| 2312 |
+
"step": 76800
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 6.77,
|
| 2316 |
+
"learning_rate": 6.8672016761135115e-06,
|
| 2317 |
+
"loss": 0.0987,
|
| 2318 |
+
"step": 77000
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 6.79,
|
| 2322 |
+
"learning_rate": 6.829788428082384e-06,
|
| 2323 |
+
"loss": 0.0946,
|
| 2324 |
+
"step": 77200
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 6.81,
|
| 2328 |
+
"learning_rate": 6.792375180051257e-06,
|
| 2329 |
+
"loss": 0.0915,
|
| 2330 |
+
"step": 77400
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 6.82,
|
| 2334 |
+
"learning_rate": 6.754961932020129e-06,
|
| 2335 |
+
"loss": 0.0889,
|
| 2336 |
+
"step": 77600
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 6.84,
|
| 2340 |
+
"learning_rate": 6.717548683989002e-06,
|
| 2341 |
+
"loss": 0.0884,
|
| 2342 |
+
"step": 77800
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 6.86,
|
| 2346 |
+
"learning_rate": 6.680135435957873e-06,
|
| 2347 |
+
"loss": 0.0854,
|
| 2348 |
+
"step": 78000
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 6.88,
|
| 2352 |
+
"learning_rate": 6.642722187926746e-06,
|
| 2353 |
+
"loss": 0.0847,
|
| 2354 |
+
"step": 78200
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 6.89,
|
| 2358 |
+
"learning_rate": 6.6053089398956175e-06,
|
| 2359 |
+
"loss": 0.0907,
|
| 2360 |
+
"step": 78400
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 6.91,
|
| 2364 |
+
"learning_rate": 6.5678956918644894e-06,
|
| 2365 |
+
"loss": 0.0955,
|
| 2366 |
+
"step": 78600
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 6.93,
|
| 2370 |
+
"learning_rate": 6.530482443833361e-06,
|
| 2371 |
+
"loss": 0.095,
|
| 2372 |
+
"step": 78800
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 6.95,
|
| 2376 |
+
"learning_rate": 6.493069195802234e-06,
|
| 2377 |
+
"loss": 0.0948,
|
| 2378 |
+
"step": 79000
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 6.97,
|
| 2382 |
+
"learning_rate": 6.455655947771107e-06,
|
| 2383 |
+
"loss": 0.0774,
|
| 2384 |
+
"step": 79200
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 6.98,
|
| 2388 |
+
"learning_rate": 6.418242699739978e-06,
|
| 2389 |
+
"loss": 0.0973,
|
| 2390 |
+
"step": 79400
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 7.0,
|
| 2394 |
+
"learning_rate": 6.380829451708851e-06,
|
| 2395 |
+
"loss": 0.0976,
|
| 2396 |
+
"step": 79600
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 7.02,
|
| 2400 |
+
"learning_rate": 6.343416203677723e-06,
|
| 2401 |
+
"loss": 0.0674,
|
| 2402 |
+
"step": 79800
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 7.04,
|
| 2406 |
+
"learning_rate": 6.306002955646595e-06,
|
| 2407 |
+
"loss": 0.0735,
|
| 2408 |
+
"step": 80000
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 7.05,
|
| 2412 |
+
"learning_rate": 6.2685897076154665e-06,
|
| 2413 |
+
"loss": 0.0777,
|
| 2414 |
+
"step": 80200
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 7.07,
|
| 2418 |
+
"learning_rate": 6.231176459584339e-06,
|
| 2419 |
+
"loss": 0.0688,
|
| 2420 |
+
"step": 80400
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 7.09,
|
| 2424 |
+
"learning_rate": 6.193763211553212e-06,
|
| 2425 |
+
"loss": 0.0721,
|
| 2426 |
+
"step": 80600
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 7.11,
|
| 2430 |
+
"learning_rate": 6.156349963522084e-06,
|
| 2431 |
+
"loss": 0.0787,
|
| 2432 |
+
"step": 80800
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 7.12,
|
| 2436 |
+
"learning_rate": 6.118936715490956e-06,
|
| 2437 |
+
"loss": 0.0755,
|
| 2438 |
+
"step": 81000
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 7.14,
|
| 2442 |
+
"learning_rate": 6.081523467459828e-06,
|
| 2443 |
+
"loss": 0.072,
|
| 2444 |
+
"step": 81200
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 7.16,
|
| 2448 |
+
"learning_rate": 6.0441102194287005e-06,
|
| 2449 |
+
"loss": 0.0695,
|
| 2450 |
+
"step": 81400
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 7.18,
|
| 2454 |
+
"learning_rate": 6.0066969713975724e-06,
|
| 2455 |
+
"loss": 0.0713,
|
| 2456 |
+
"step": 81600
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 7.19,
|
| 2460 |
+
"learning_rate": 5.969283723366444e-06,
|
| 2461 |
+
"loss": 0.0771,
|
| 2462 |
+
"step": 81800
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 7.21,
|
| 2466 |
+
"learning_rate": 5.931870475335316e-06,
|
| 2467 |
+
"loss": 0.0695,
|
| 2468 |
+
"step": 82000
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 7.23,
|
| 2472 |
+
"learning_rate": 5.894457227304189e-06,
|
| 2473 |
+
"loss": 0.0676,
|
| 2474 |
+
"step": 82200
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 7.25,
|
| 2478 |
+
"learning_rate": 5.857043979273062e-06,
|
| 2479 |
+
"loss": 0.0766,
|
| 2480 |
+
"step": 82400
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 7.26,
|
| 2484 |
+
"learning_rate": 5.819630731241933e-06,
|
| 2485 |
+
"loss": 0.0766,
|
| 2486 |
+
"step": 82600
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 7.28,
|
| 2490 |
+
"learning_rate": 5.782217483210806e-06,
|
| 2491 |
+
"loss": 0.0807,
|
| 2492 |
+
"step": 82800
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 7.3,
|
| 2496 |
+
"learning_rate": 5.7448042351796775e-06,
|
| 2497 |
+
"loss": 0.0834,
|
| 2498 |
+
"step": 83000
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 7.32,
|
| 2502 |
+
"learning_rate": 5.70739098714855e-06,
|
| 2503 |
+
"loss": 0.0789,
|
| 2504 |
+
"step": 83200
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 7.33,
|
| 2508 |
+
"learning_rate": 5.669977739117421e-06,
|
| 2509 |
+
"loss": 0.0691,
|
| 2510 |
+
"step": 83400
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 7.35,
|
| 2514 |
+
"learning_rate": 5.632564491086294e-06,
|
| 2515 |
+
"loss": 0.0777,
|
| 2516 |
+
"step": 83600
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 7.37,
|
| 2520 |
+
"learning_rate": 5.595151243055166e-06,
|
| 2521 |
+
"loss": 0.0725,
|
| 2522 |
+
"step": 83800
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 7.39,
|
| 2526 |
+
"learning_rate": 5.557737995024039e-06,
|
| 2527 |
+
"loss": 0.0788,
|
| 2528 |
+
"step": 84000
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 7.4,
|
| 2532 |
+
"learning_rate": 5.520324746992911e-06,
|
| 2533 |
+
"loss": 0.075,
|
| 2534 |
+
"step": 84200
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 7.42,
|
| 2538 |
+
"learning_rate": 5.482911498961783e-06,
|
| 2539 |
+
"loss": 0.0742,
|
| 2540 |
+
"step": 84400
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 7.44,
|
| 2544 |
+
"learning_rate": 5.4454982509306554e-06,
|
| 2545 |
+
"loss": 0.0666,
|
| 2546 |
+
"step": 84600
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 7.46,
|
| 2550 |
+
"learning_rate": 5.4080850028995265e-06,
|
| 2551 |
+
"loss": 0.0688,
|
| 2552 |
+
"step": 84800
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 7.48,
|
| 2556 |
+
"learning_rate": 5.370671754868399e-06,
|
| 2557 |
+
"loss": 0.0747,
|
| 2558 |
+
"step": 85000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 7.49,
|
| 2562 |
+
"learning_rate": 5.333258506837271e-06,
|
| 2563 |
+
"loss": 0.0741,
|
| 2564 |
+
"step": 85200
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 7.51,
|
| 2568 |
+
"learning_rate": 5.295845258806144e-06,
|
| 2569 |
+
"loss": 0.0657,
|
| 2570 |
+
"step": 85400
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 7.53,
|
| 2574 |
+
"learning_rate": 5.258432010775017e-06,
|
| 2575 |
+
"loss": 0.0788,
|
| 2576 |
+
"step": 85600
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 7.55,
|
| 2580 |
+
"learning_rate": 5.221018762743888e-06,
|
| 2581 |
+
"loss": 0.0791,
|
| 2582 |
+
"step": 85800
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 7.56,
|
| 2586 |
+
"learning_rate": 5.1836055147127605e-06,
|
| 2587 |
+
"loss": 0.0752,
|
| 2588 |
+
"step": 86000
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 7.58,
|
| 2592 |
+
"learning_rate": 5.1461922666816325e-06,
|
| 2593 |
+
"loss": 0.0762,
|
| 2594 |
+
"step": 86200
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 7.6,
|
| 2598 |
+
"learning_rate": 5.108779018650505e-06,
|
| 2599 |
+
"loss": 0.0771,
|
| 2600 |
+
"step": 86400
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 7.62,
|
| 2604 |
+
"learning_rate": 5.071365770619376e-06,
|
| 2605 |
+
"loss": 0.0727,
|
| 2606 |
+
"step": 86600
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 7.63,
|
| 2610 |
+
"learning_rate": 5.033952522588249e-06,
|
| 2611 |
+
"loss": 0.0843,
|
| 2612 |
+
"step": 86800
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 7.65,
|
| 2616 |
+
"learning_rate": 4.996539274557121e-06,
|
| 2617 |
+
"loss": 0.0741,
|
| 2618 |
+
"step": 87000
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 7.67,
|
| 2622 |
+
"learning_rate": 4.959126026525993e-06,
|
| 2623 |
+
"loss": 0.0721,
|
| 2624 |
+
"step": 87200
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 7.69,
|
| 2628 |
+
"learning_rate": 4.921712778494866e-06,
|
| 2629 |
+
"loss": 0.0814,
|
| 2630 |
+
"step": 87400
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 7.7,
|
| 2634 |
+
"learning_rate": 4.884299530463738e-06,
|
| 2635 |
+
"loss": 0.0703,
|
| 2636 |
+
"step": 87600
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 7.72,
|
| 2640 |
+
"learning_rate": 4.8468862824326095e-06,
|
| 2641 |
+
"loss": 0.0745,
|
| 2642 |
+
"step": 87800
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 7.74,
|
| 2646 |
+
"learning_rate": 4.809473034401481e-06,
|
| 2647 |
+
"loss": 0.0788,
|
| 2648 |
+
"step": 88000
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 7.76,
|
| 2652 |
+
"learning_rate": 4.772059786370354e-06,
|
| 2653 |
+
"loss": 0.0721,
|
| 2654 |
+
"step": 88200
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 7.77,
|
| 2658 |
+
"learning_rate": 4.734646538339226e-06,
|
| 2659 |
+
"loss": 0.0689,
|
| 2660 |
+
"step": 88400
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 7.79,
|
| 2664 |
+
"learning_rate": 4.697233290308099e-06,
|
| 2665 |
+
"loss": 0.0651,
|
| 2666 |
+
"step": 88600
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 7.81,
|
| 2670 |
+
"learning_rate": 4.659820042276971e-06,
|
| 2671 |
+
"loss": 0.0775,
|
| 2672 |
+
"step": 88800
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 7.83,
|
| 2676 |
+
"learning_rate": 4.622406794245843e-06,
|
| 2677 |
+
"loss": 0.069,
|
| 2678 |
+
"step": 89000
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 7.84,
|
| 2682 |
+
"learning_rate": 4.584993546214715e-06,
|
| 2683 |
+
"loss": 0.0807,
|
| 2684 |
+
"step": 89200
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 7.86,
|
| 2688 |
+
"learning_rate": 4.547580298183587e-06,
|
| 2689 |
+
"loss": 0.0817,
|
| 2690 |
+
"step": 89400
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 7.88,
|
| 2694 |
+
"learning_rate": 4.510167050152459e-06,
|
| 2695 |
+
"loss": 0.0771,
|
| 2696 |
+
"step": 89600
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 7.9,
|
| 2700 |
+
"learning_rate": 4.472753802121331e-06,
|
| 2701 |
+
"loss": 0.0683,
|
| 2702 |
+
"step": 89800
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 7.91,
|
| 2706 |
+
"learning_rate": 4.435340554090204e-06,
|
| 2707 |
+
"loss": 0.0704,
|
| 2708 |
+
"step": 90000
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 7.93,
|
| 2712 |
+
"learning_rate": 4.397927306059076e-06,
|
| 2713 |
+
"loss": 0.0852,
|
| 2714 |
+
"step": 90200
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 7.95,
|
| 2718 |
+
"learning_rate": 4.360514058027948e-06,
|
| 2719 |
+
"loss": 0.0773,
|
| 2720 |
+
"step": 90400
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 7.97,
|
| 2724 |
+
"learning_rate": 4.323100809996821e-06,
|
| 2725 |
+
"loss": 0.0694,
|
| 2726 |
+
"step": 90600
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 7.99,
|
| 2730 |
+
"learning_rate": 4.2856875619656925e-06,
|
| 2731 |
+
"loss": 0.0771,
|
| 2732 |
+
"step": 90800
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 8.0,
|
| 2736 |
+
"learning_rate": 4.2482743139345644e-06,
|
| 2737 |
+
"loss": 0.0702,
|
| 2738 |
+
"step": 91000
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 8.02,
|
| 2742 |
+
"learning_rate": 4.210861065903436e-06,
|
| 2743 |
+
"loss": 0.0533,
|
| 2744 |
+
"step": 91200
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 8.04,
|
| 2748 |
+
"learning_rate": 4.173447817872308e-06,
|
| 2749 |
+
"loss": 0.0666,
|
| 2750 |
+
"step": 91400
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 8.06,
|
| 2754 |
+
"learning_rate": 4.136034569841181e-06,
|
| 2755 |
+
"loss": 0.0566,
|
| 2756 |
+
"step": 91600
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 8.07,
|
| 2760 |
+
"learning_rate": 4.098621321810054e-06,
|
| 2761 |
+
"loss": 0.0611,
|
| 2762 |
+
"step": 91800
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 8.09,
|
| 2766 |
+
"learning_rate": 4.061208073778926e-06,
|
| 2767 |
+
"loss": 0.0589,
|
| 2768 |
+
"step": 92000
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 8.11,
|
| 2772 |
+
"learning_rate": 4.023794825747798e-06,
|
| 2773 |
+
"loss": 0.0602,
|
| 2774 |
+
"step": 92200
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 8.13,
|
| 2778 |
+
"learning_rate": 3.9863815777166695e-06,
|
| 2779 |
+
"loss": 0.0615,
|
| 2780 |
+
"step": 92400
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 8.14,
|
| 2784 |
+
"learning_rate": 3.948968329685542e-06,
|
| 2785 |
+
"loss": 0.0682,
|
| 2786 |
+
"step": 92600
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 8.16,
|
| 2790 |
+
"learning_rate": 3.911555081654414e-06,
|
| 2791 |
+
"loss": 0.0678,
|
| 2792 |
+
"step": 92800
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 8.18,
|
| 2796 |
+
"learning_rate": 3.874141833623286e-06,
|
| 2797 |
+
"loss": 0.0586,
|
| 2798 |
+
"step": 93000
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 8.2,
|
| 2802 |
+
"learning_rate": 3.836728585592159e-06,
|
| 2803 |
+
"loss": 0.0641,
|
| 2804 |
+
"step": 93200
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 8.21,
|
| 2808 |
+
"learning_rate": 3.799315337561031e-06,
|
| 2809 |
+
"loss": 0.0586,
|
| 2810 |
+
"step": 93400
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 8.23,
|
| 2814 |
+
"learning_rate": 3.761902089529903e-06,
|
| 2815 |
+
"loss": 0.0684,
|
| 2816 |
+
"step": 93600
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 8.25,
|
| 2820 |
+
"learning_rate": 3.724488841498775e-06,
|
| 2821 |
+
"loss": 0.0594,
|
| 2822 |
+
"step": 93800
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 8.27,
|
| 2826 |
+
"learning_rate": 3.687075593467647e-06,
|
| 2827 |
+
"loss": 0.061,
|
| 2828 |
+
"step": 94000
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 8.28,
|
| 2832 |
+
"learning_rate": 3.6496623454365193e-06,
|
| 2833 |
+
"loss": 0.0587,
|
| 2834 |
+
"step": 94200
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 8.3,
|
| 2838 |
+
"learning_rate": 3.6122490974053913e-06,
|
| 2839 |
+
"loss": 0.0675,
|
| 2840 |
+
"step": 94400
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 8.32,
|
| 2844 |
+
"learning_rate": 3.5748358493742636e-06,
|
| 2845 |
+
"loss": 0.0621,
|
| 2846 |
+
"step": 94600
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 8.34,
|
| 2850 |
+
"learning_rate": 3.5374226013431355e-06,
|
| 2851 |
+
"loss": 0.0556,
|
| 2852 |
+
"step": 94800
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 8.35,
|
| 2856 |
+
"learning_rate": 3.5000093533120083e-06,
|
| 2857 |
+
"loss": 0.0728,
|
| 2858 |
+
"step": 95000
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 8.37,
|
| 2862 |
+
"learning_rate": 3.46259610528088e-06,
|
| 2863 |
+
"loss": 0.0554,
|
| 2864 |
+
"step": 95200
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 8.39,
|
| 2868 |
+
"learning_rate": 3.4251828572497525e-06,
|
| 2869 |
+
"loss": 0.0634,
|
| 2870 |
+
"step": 95400
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 8.41,
|
| 2874 |
+
"learning_rate": 3.3877696092186245e-06,
|
| 2875 |
+
"loss": 0.063,
|
| 2876 |
+
"step": 95600
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 8.42,
|
| 2880 |
+
"learning_rate": 3.350356361187497e-06,
|
| 2881 |
+
"loss": 0.0613,
|
| 2882 |
+
"step": 95800
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 8.44,
|
| 2886 |
+
"learning_rate": 3.3129431131563687e-06,
|
| 2887 |
+
"loss": 0.0755,
|
| 2888 |
+
"step": 96000
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 8.46,
|
| 2892 |
+
"learning_rate": 3.275529865125241e-06,
|
| 2893 |
+
"loss": 0.0616,
|
| 2894 |
+
"step": 96200
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 8.48,
|
| 2898 |
+
"learning_rate": 3.238116617094113e-06,
|
| 2899 |
+
"loss": 0.0513,
|
| 2900 |
+
"step": 96400
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 8.5,
|
| 2904 |
+
"learning_rate": 3.2007033690629857e-06,
|
| 2905 |
+
"loss": 0.0609,
|
| 2906 |
+
"step": 96600
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 8.51,
|
| 2910 |
+
"learning_rate": 3.1632901210318577e-06,
|
| 2911 |
+
"loss": 0.0625,
|
| 2912 |
+
"step": 96800
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 8.53,
|
| 2916 |
+
"learning_rate": 3.12587687300073e-06,
|
| 2917 |
+
"loss": 0.057,
|
| 2918 |
+
"step": 97000
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 8.55,
|
| 2922 |
+
"learning_rate": 3.088463624969602e-06,
|
| 2923 |
+
"loss": 0.0655,
|
| 2924 |
+
"step": 97200
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 8.57,
|
| 2928 |
+
"learning_rate": 3.0510503769384743e-06,
|
| 2929 |
+
"loss": 0.0543,
|
| 2930 |
+
"step": 97400
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 8.58,
|
| 2934 |
+
"learning_rate": 3.013637128907346e-06,
|
| 2935 |
+
"loss": 0.077,
|
| 2936 |
+
"step": 97600
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 8.6,
|
| 2940 |
+
"learning_rate": 2.9762238808762185e-06,
|
| 2941 |
+
"loss": 0.0487,
|
| 2942 |
+
"step": 97800
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 8.62,
|
| 2946 |
+
"learning_rate": 2.9388106328450904e-06,
|
| 2947 |
+
"loss": 0.0655,
|
| 2948 |
+
"step": 98000
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 8.64,
|
| 2952 |
+
"learning_rate": 2.9013973848139628e-06,
|
| 2953 |
+
"loss": 0.0655,
|
| 2954 |
+
"step": 98200
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 8.65,
|
| 2958 |
+
"learning_rate": 2.863984136782835e-06,
|
| 2959 |
+
"loss": 0.0596,
|
| 2960 |
+
"step": 98400
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 8.67,
|
| 2964 |
+
"learning_rate": 2.8265708887517075e-06,
|
| 2965 |
+
"loss": 0.0594,
|
| 2966 |
+
"step": 98600
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 8.69,
|
| 2970 |
+
"learning_rate": 2.7891576407205794e-06,
|
| 2971 |
+
"loss": 0.0737,
|
| 2972 |
+
"step": 98800
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 8.71,
|
| 2976 |
+
"learning_rate": 2.7517443926894517e-06,
|
| 2977 |
+
"loss": 0.0616,
|
| 2978 |
+
"step": 99000
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 8.72,
|
| 2982 |
+
"learning_rate": 2.7143311446583236e-06,
|
| 2983 |
+
"loss": 0.0531,
|
| 2984 |
+
"step": 99200
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 8.74,
|
| 2988 |
+
"learning_rate": 2.676917896627196e-06,
|
| 2989 |
+
"loss": 0.0717,
|
| 2990 |
+
"step": 99400
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 8.76,
|
| 2994 |
+
"learning_rate": 2.639504648596068e-06,
|
| 2995 |
+
"loss": 0.0643,
|
| 2996 |
+
"step": 99600
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 8.78,
|
| 3000 |
+
"learning_rate": 2.60209140056494e-06,
|
| 3001 |
+
"loss": 0.0515,
|
| 3002 |
+
"step": 99800
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 8.79,
|
| 3006 |
+
"learning_rate": 2.5646781525338126e-06,
|
| 3007 |
+
"loss": 0.0612,
|
| 3008 |
+
"step": 100000
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 8.81,
|
| 3012 |
+
"learning_rate": 2.527264904502685e-06,
|
| 3013 |
+
"loss": 0.0562,
|
| 3014 |
+
"step": 100200
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 8.83,
|
| 3018 |
+
"learning_rate": 2.489851656471557e-06,
|
| 3019 |
+
"loss": 0.0605,
|
| 3020 |
+
"step": 100400
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 8.85,
|
| 3024 |
+
"learning_rate": 2.452438408440429e-06,
|
| 3025 |
+
"loss": 0.0621,
|
| 3026 |
+
"step": 100600
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 8.86,
|
| 3030 |
+
"learning_rate": 2.415025160409301e-06,
|
| 3031 |
+
"loss": 0.0639,
|
| 3032 |
+
"step": 100800
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 8.88,
|
| 3036 |
+
"learning_rate": 2.3776119123781734e-06,
|
| 3037 |
+
"loss": 0.0565,
|
| 3038 |
+
"step": 101000
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 8.9,
|
| 3042 |
+
"learning_rate": 2.3401986643470454e-06,
|
| 3043 |
+
"loss": 0.0609,
|
| 3044 |
+
"step": 101200
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 8.92,
|
| 3048 |
+
"learning_rate": 2.3027854163159177e-06,
|
| 3049 |
+
"loss": 0.0613,
|
| 3050 |
+
"step": 101400
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 8.94,
|
| 3054 |
+
"learning_rate": 2.26537216828479e-06,
|
| 3055 |
+
"loss": 0.0673,
|
| 3056 |
+
"step": 101600
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 8.95,
|
| 3060 |
+
"learning_rate": 2.227958920253662e-06,
|
| 3061 |
+
"loss": 0.0598,
|
| 3062 |
+
"step": 101800
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 8.97,
|
| 3066 |
+
"learning_rate": 2.190545672222534e-06,
|
| 3067 |
+
"loss": 0.0649,
|
| 3068 |
+
"step": 102000
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 8.99,
|
| 3072 |
+
"learning_rate": 2.1531324241914066e-06,
|
| 3073 |
+
"loss": 0.0615,
|
| 3074 |
+
"step": 102200
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 9.01,
|
| 3078 |
+
"learning_rate": 2.1157191761602786e-06,
|
| 3079 |
+
"loss": 0.0538,
|
| 3080 |
+
"step": 102400
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 9.02,
|
| 3084 |
+
"learning_rate": 2.0783059281291505e-06,
|
| 3085 |
+
"loss": 0.0462,
|
| 3086 |
+
"step": 102600
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 9.04,
|
| 3090 |
+
"learning_rate": 2.040892680098023e-06,
|
| 3091 |
+
"loss": 0.0542,
|
| 3092 |
+
"step": 102800
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 9.06,
|
| 3096 |
+
"learning_rate": 2.003479432066895e-06,
|
| 3097 |
+
"loss": 0.0529,
|
| 3098 |
+
"step": 103000
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 9.08,
|
| 3102 |
+
"learning_rate": 1.966066184035767e-06,
|
| 3103 |
+
"loss": 0.0585,
|
| 3104 |
+
"step": 103200
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 9.09,
|
| 3108 |
+
"learning_rate": 1.9286529360046394e-06,
|
| 3109 |
+
"loss": 0.0462,
|
| 3110 |
+
"step": 103400
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 9.11,
|
| 3114 |
+
"learning_rate": 1.8912396879735116e-06,
|
| 3115 |
+
"loss": 0.0496,
|
| 3116 |
+
"step": 103600
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 9.13,
|
| 3120 |
+
"learning_rate": 1.8538264399423839e-06,
|
| 3121 |
+
"loss": 0.0477,
|
| 3122 |
+
"step": 103800
|
| 3123 |
+
},
|
| 3124 |
+
{
|
| 3125 |
+
"epoch": 9.15,
|
| 3126 |
+
"learning_rate": 1.816413191911256e-06,
|
| 3127 |
+
"loss": 0.048,
|
| 3128 |
+
"step": 104000
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 9.16,
|
| 3132 |
+
"learning_rate": 1.7789999438801282e-06,
|
| 3133 |
+
"loss": 0.0591,
|
| 3134 |
+
"step": 104200
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 9.18,
|
| 3138 |
+
"learning_rate": 1.7415866958490003e-06,
|
| 3139 |
+
"loss": 0.0445,
|
| 3140 |
+
"step": 104400
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 9.2,
|
| 3144 |
+
"learning_rate": 1.7041734478178726e-06,
|
| 3145 |
+
"loss": 0.0508,
|
| 3146 |
+
"step": 104600
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 9.22,
|
| 3150 |
+
"learning_rate": 1.6667601997867448e-06,
|
| 3151 |
+
"loss": 0.0534,
|
| 3152 |
+
"step": 104800
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 9.23,
|
| 3156 |
+
"learning_rate": 1.6293469517556169e-06,
|
| 3157 |
+
"loss": 0.046,
|
| 3158 |
+
"step": 105000
|
| 3159 |
+
},
|
| 3160 |
+
{
|
| 3161 |
+
"epoch": 9.25,
|
| 3162 |
+
"learning_rate": 1.591933703724489e-06,
|
| 3163 |
+
"loss": 0.0454,
|
| 3164 |
+
"step": 105200
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 9.27,
|
| 3168 |
+
"learning_rate": 1.554520455693361e-06,
|
| 3169 |
+
"loss": 0.0601,
|
| 3170 |
+
"step": 105400
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 9.29,
|
| 3174 |
+
"learning_rate": 1.5171072076622335e-06,
|
| 3175 |
+
"loss": 0.0543,
|
| 3176 |
+
"step": 105600
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 9.3,
|
| 3180 |
+
"learning_rate": 1.4796939596311056e-06,
|
| 3181 |
+
"loss": 0.0587,
|
| 3182 |
+
"step": 105800
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 9.32,
|
| 3186 |
+
"learning_rate": 1.4422807115999775e-06,
|
| 3187 |
+
"loss": 0.0526,
|
| 3188 |
+
"step": 106000
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 9.34,
|
| 3192 |
+
"learning_rate": 1.4048674635688497e-06,
|
| 3193 |
+
"loss": 0.065,
|
| 3194 |
+
"step": 106200
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 9.36,
|
| 3198 |
+
"learning_rate": 1.3674542155377222e-06,
|
| 3199 |
+
"loss": 0.0531,
|
| 3200 |
+
"step": 106400
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 9.37,
|
| 3204 |
+
"learning_rate": 1.3300409675065941e-06,
|
| 3205 |
+
"loss": 0.0607,
|
| 3206 |
+
"step": 106600
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 9.39,
|
| 3210 |
+
"learning_rate": 1.2926277194754663e-06,
|
| 3211 |
+
"loss": 0.0425,
|
| 3212 |
+
"step": 106800
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 9.41,
|
| 3216 |
+
"learning_rate": 1.2552144714443384e-06,
|
| 3217 |
+
"loss": 0.0541,
|
| 3218 |
+
"step": 107000
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 9.43,
|
| 3222 |
+
"learning_rate": 1.2178012234132107e-06,
|
| 3223 |
+
"loss": 0.0592,
|
| 3224 |
+
"step": 107200
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 9.45,
|
| 3228 |
+
"learning_rate": 1.1803879753820829e-06,
|
| 3229 |
+
"loss": 0.0494,
|
| 3230 |
+
"step": 107400
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 9.46,
|
| 3234 |
+
"learning_rate": 1.142974727350955e-06,
|
| 3235 |
+
"loss": 0.0548,
|
| 3236 |
+
"step": 107600
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 9.48,
|
| 3240 |
+
"learning_rate": 1.1055614793198273e-06,
|
| 3241 |
+
"loss": 0.0439,
|
| 3242 |
+
"step": 107800
|
| 3243 |
+
},
|
| 3244 |
+
{
|
| 3245 |
+
"epoch": 9.5,
|
| 3246 |
+
"learning_rate": 1.0681482312886995e-06,
|
| 3247 |
+
"loss": 0.0543,
|
| 3248 |
+
"step": 108000
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 9.52,
|
| 3252 |
+
"learning_rate": 1.0307349832575716e-06,
|
| 3253 |
+
"loss": 0.0604,
|
| 3254 |
+
"step": 108200
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 9.53,
|
| 3258 |
+
"learning_rate": 9.933217352264437e-07,
|
| 3259 |
+
"loss": 0.0546,
|
| 3260 |
+
"step": 108400
|
| 3261 |
+
},
|
| 3262 |
+
{
|
| 3263 |
+
"epoch": 9.55,
|
| 3264 |
+
"learning_rate": 9.559084871953159e-07,
|
| 3265 |
+
"loss": 0.0576,
|
| 3266 |
+
"step": 108600
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 9.57,
|
| 3270 |
+
"learning_rate": 9.184952391641881e-07,
|
| 3271 |
+
"loss": 0.0441,
|
| 3272 |
+
"step": 108800
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 9.59,
|
| 3276 |
+
"learning_rate": 8.810819911330602e-07,
|
| 3277 |
+
"loss": 0.05,
|
| 3278 |
+
"step": 109000
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 9.6,
|
| 3282 |
+
"learning_rate": 8.436687431019325e-07,
|
| 3283 |
+
"loss": 0.0523,
|
| 3284 |
+
"step": 109200
|
| 3285 |
+
},
|
| 3286 |
+
{
|
| 3287 |
+
"epoch": 9.62,
|
| 3288 |
+
"learning_rate": 8.062554950708046e-07,
|
| 3289 |
+
"loss": 0.053,
|
| 3290 |
+
"step": 109400
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 9.64,
|
| 3294 |
+
"learning_rate": 7.688422470396768e-07,
|
| 3295 |
+
"loss": 0.0456,
|
| 3296 |
+
"step": 109600
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 9.66,
|
| 3300 |
+
"learning_rate": 7.314289990085489e-07,
|
| 3301 |
+
"loss": 0.0508,
|
| 3302 |
+
"step": 109800
|
| 3303 |
+
},
|
| 3304 |
+
{
|
| 3305 |
+
"epoch": 9.67,
|
| 3306 |
+
"learning_rate": 6.940157509774212e-07,
|
| 3307 |
+
"loss": 0.0415,
|
| 3308 |
+
"step": 110000
|
| 3309 |
+
}
|
| 3310 |
+
],
|
| 3311 |
+
"max_steps": 113710,
|
| 3312 |
+
"num_train_epochs": 10,
|
| 3313 |
+
"total_flos": 2.6986684097812992e+17,
|
| 3314 |
+
"trial_name": null,
|
| 3315 |
+
"trial_params": null
|
| 3316 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:deb6d415dd7293fb9fc7ecae0a9a429f94eed07075dd532bf210eeb896bec7a9
|
| 3 |
+
size 2991
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|